# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details. from __future__ import annotations import inspect import warnings from typing import TYPE_CHECKING, List, Type, Union, Iterable, Optional, cast from functools import partial from itertools import chain from typing_extensions import Literal, overload import httpx import pydantic from .... import _legacy_response from .batches import ( Batches, AsyncBatches, BatchesWithRawResponse, AsyncBatchesWithRawResponse, BatchesWithStreamingResponse, AsyncBatchesWithStreamingResponse, ) from ...._types import NOT_GIVEN, Body, Omit, Query, Headers, NotGiven, SequenceNotStr, omit, not_given from ...._utils import is_given, required_args, maybe_transform, strip_not_given, async_maybe_transform from ...._compat import cached_property from ...._models import TypeAdapter from ...._resource import SyncAPIResource, AsyncAPIResource from ...._response import to_streamed_response_wrapper, async_to_streamed_response_wrapper from ....lib.tools import ( BetaToolRunner, BetaAsyncToolRunner, BetaStreamingToolRunner, BetaAsyncStreamingToolRunner, ) from ...._constants import DEFAULT_TIMEOUT, MODEL_NONSTREAMING_TOKENS from ...._streaming import Stream, AsyncStream from ....types.beta import ( BetaThinkingConfigParam, message_create_params, message_count_tokens_params, ) from ...._base_client import make_request_options from ....lib.streaming import BetaMessageStreamManager, BetaAsyncMessageStreamManager from ...messages.messages import DEPRECATED_MODELS from ....types.model_param import ModelParam from ....lib._parse._response import ResponseFormatT, parse_response from ....lib._parse._transform import transform_schema from ....types.beta.beta_message import BetaMessage from ....lib.tools._beta_functions import BetaRunnableTool, BetaAsyncRunnableTool from ....types.anthropic_beta_param import AnthropicBetaParam from ....types.beta.beta_message_param import BetaMessageParam from ....types.beta.beta_metadata_param import BetaMetadataParam from ....types.beta.parsed_beta_message import ParsedBetaMessage from ....types.beta.beta_text_block_param import BetaTextBlockParam from ....types.beta.beta_tool_union_param import BetaToolUnionParam from ....types.beta.beta_tool_choice_param import BetaToolChoiceParam from ....lib.tools._beta_compaction_control import CompactionControl from ....types.beta.beta_output_config_param import BetaOutputConfigParam from ....types.beta.beta_message_tokens_count import BetaMessageTokensCount from ....types.beta.beta_thinking_config_param import BetaThinkingConfigParam from ....types.beta.beta_json_output_format_param import BetaJSONOutputFormatParam from ....types.beta.beta_raw_message_stream_event import BetaRawMessageStreamEvent from ....types.beta.beta_context_management_config_param import BetaContextManagementConfigParam from ....types.beta.beta_request_mcp_server_url_definition_param import BetaRequestMCPServerURLDefinitionParam if TYPE_CHECKING: from ...._client import Anthropic, AsyncAnthropic __all__ = ["Messages", "AsyncMessages"] class Messages(SyncAPIResource): @cached_property def batches(self) -> Batches: return Batches(self._client) @cached_property def with_raw_response(self) -> MessagesWithRawResponse: """ This property can be used as a prefix for any HTTP method call to return the raw response object instead of the parsed content. For more information, see https://www.github.com/anthropics/anthropic-sdk-python#accessing-raw-response-data-eg-headers """ return MessagesWithRawResponse(self) @cached_property def with_streaming_response(self) -> MessagesWithStreamingResponse: """ An alternative to `.with_raw_response` that doesn't eagerly read the response body. For more information, see https://www.github.com/anthropics/anthropic-sdk-python#with_streaming_response """ return MessagesWithStreamingResponse(self) @overload def create( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: Literal[False] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> BetaMessage: """ Send a structured list of input messages with text and/or image content, and the model will generate the next message in the conversation. The Messages API can be used for either single queries or stateless multi-turn conversations. Learn more about the Messages API in our [user guide](https://docs.claude.com/en/docs/initial-setup) Args: max_tokens: The maximum number of tokens to generate before stopping. Note that our models may stop _before_ reaching this maximum. This parameter only specifies the absolute maximum number of tokens to generate. Different models have different maximum values for this parameter. See [models](https://docs.claude.com/en/docs/models-overview) for details. messages: Input messages. Our models are trained to operate on alternating `user` and `assistant` conversational turns. When creating a new `Message`, you specify the prior conversational turns with the `messages` parameter, and the model then generates the next `Message` in the conversation. Consecutive `user` or `assistant` turns in your request will be combined into a single turn. Each input message must be an object with a `role` and `content`. You can specify a single `user`-role message, or you can include multiple `user` and `assistant` messages. If the final message uses the `assistant` role, the response content will continue immediately from the content in that message. This can be used to constrain part of the model's response. Example with a single `user` message: ```json [{ "role": "user", "content": "Hello, Claude" }] ``` Example with multiple conversational turns: ```json [ { "role": "user", "content": "Hello there." }, { "role": "assistant", "content": "Hi, I'm Claude. How can I help you?" }, { "role": "user", "content": "Can you explain LLMs in plain English?" } ] ``` Example with a partially-filled response from Claude: ```json [ { "role": "user", "content": "What's the Greek name for Sun? (A) Sol (B) Helios (C) Sun" }, { "role": "assistant", "content": "The best answer is (" } ] ``` Each input message `content` may be either a single `string` or an array of content blocks, where each block has a specific `type`. Using a `string` for `content` is shorthand for an array of one content block of type `"text"`. The following input messages are equivalent: ```json { "role": "user", "content": "Hello, Claude" } ``` ```json { "role": "user", "content": [{ "type": "text", "text": "Hello, Claude" }] } ``` See [input examples](https://docs.claude.com/en/api/messages-examples). Note that if you want to include a [system prompt](https://docs.claude.com/en/docs/system-prompts), you can use the top-level `system` parameter — there is no `"system"` role for input messages in the Messages API. There is a limit of 100,000 messages in a single request. model: The model that will complete your prompt.\n\nSee [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. container: Container identifier for reuse across requests. context_management: Context management configuration. This allows you to control how Claude manages context across multiple requests, such as whether to clear function results or not. mcp_servers: MCP servers to be utilized in this request metadata: An object describing metadata about the request. output_config: Configuration options for the model's output. Controls aspects like how much effort the model puts into its response. output_format: A schema to specify Claude's output format in responses. service_tier: Determines whether to use priority capacity (if available) or standard capacity for this request. Anthropic offers different levels of service for your API requests. See [service-tiers](https://docs.claude.com/en/api/service-tiers) for details. stop_sequences: Custom text sequences that will cause the model to stop generating. Our models will normally stop when they have naturally completed their turn, which will result in a response `stop_reason` of `"end_turn"`. If you want the model to stop generating when it encounters custom strings of text, you can use the `stop_sequences` parameter. If the model encounters one of the custom sequences, the response `stop_reason` value will be `"stop_sequence"` and the response `stop_sequence` value will contain the matched stop sequence. stream: Whether to incrementally stream the response using server-sent events. See [streaming](https://docs.claude.com/en/api/messages-streaming) for details. system: System prompt. A system prompt is a way of providing context and instructions to Claude, such as specifying a particular goal or role. See our [guide to system prompts](https://docs.claude.com/en/docs/system-prompts). temperature: Amount of randomness injected into the response. Defaults to `1.0`. Ranges from `0.0` to `1.0`. Use `temperature` closer to `0.0` for analytical / multiple choice, and closer to `1.0` for creative and generative tasks. Note that even with `temperature` of `0.0`, the results will not be fully deterministic. thinking: Configuration for enabling Claude's extended thinking. When enabled, responses include `thinking` content blocks showing Claude's thinking process before the final answer. Requires a minimum budget of 1,024 tokens and counts towards your `max_tokens` limit. See [extended thinking](https://docs.claude.com/en/docs/build-with-claude/extended-thinking) for details. tool_choice: How the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all. tools: Definitions of tools that the model may use. If you include `tools` in your API request, the model may return `tool_use` content blocks that represent the model's use of those tools. You can then run those tools using the tool input generated by the model and then optionally return results back to the model using `tool_result` content blocks. There are two types of tools: **client tools** and **server tools**. The behavior described below applies to client tools. For [server tools](https://docs.claude.com/en/docs/agents-and-tools/tool-use/overview#server-tools), see their individual documentation as each has its own behavior (e.g., the [web search tool](https://docs.claude.com/en/docs/agents-and-tools/tool-use/web-search-tool)). Each tool definition includes: - `name`: Name of the tool. - `description`: Optional, but strongly-recommended description of the tool. - `input_schema`: [JSON schema](https://json-schema.org/draft/2020-12) for the tool `input` shape that the model will produce in `tool_use` output content blocks. For example, if you defined `tools` as: ```json [ { "name": "get_stock_price", "description": "Get the current stock price for a given ticker symbol.", "input_schema": { "type": "object", "properties": { "ticker": { "type": "string", "description": "The stock ticker symbol, e.g. AAPL for Apple Inc." } }, "required": ["ticker"] } } ] ``` And then asked the model "What's the S&P 500 at today?", the model might produce `tool_use` content blocks in the response like this: ```json [ { "type": "tool_use", "id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "name": "get_stock_price", "input": { "ticker": "^GSPC" } } ] ``` You might then run your `get_stock_price` tool with `{"ticker": "^GSPC"}` as an input, and return the following back to the model in a subsequent `user` message: ```json [ { "type": "tool_result", "tool_use_id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "content": "259.75 USD" } ] ``` Tools can be used for workflows that include running client-side tools and functions, or more generally whenever you want the model to produce a particular JSON structure of output. See our [guide](https://docs.claude.com/en/docs/tool-use) for more details. top_k: Only sample from the top K options for each subsequent token. Used to remove "long tail" low probability responses. [Learn more technical details here](https://towardsdatascience.com/how-to-sample-from-language-models-682bceb97277). Recommended for advanced use cases only. You usually only need to use `temperature`. top_p: Use nucleus sampling. In nucleus sampling, we compute the cumulative distribution over all the options for each subsequent token in decreasing probability order and cut it off once it reaches a particular probability specified by `top_p`. You should either alter `temperature` or `top_p`, but not both. Recommended for advanced use cases only. You usually only need to use `temperature`. betas: Optional header to specify the beta version(s) you want to use. extra_headers: Send extra headers extra_query: Add additional query parameters to the request extra_body: Add additional JSON properties to the request timeout: Override the client-level default timeout for this request, in seconds """ ... @overload def create( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, stream: Literal[True], container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> Stream[BetaRawMessageStreamEvent]: """ Send a structured list of input messages with text and/or image content, and the model will generate the next message in the conversation. The Messages API can be used for either single queries or stateless multi-turn conversations. Learn more about the Messages API in our [user guide](https://docs.claude.com/en/docs/initial-setup) Args: max_tokens: The maximum number of tokens to generate before stopping. Note that our models may stop _before_ reaching this maximum. This parameter only specifies the absolute maximum number of tokens to generate. Different models have different maximum values for this parameter. See [models](https://docs.claude.com/en/docs/models-overview) for details. messages: Input messages. Our models are trained to operate on alternating `user` and `assistant` conversational turns. When creating a new `Message`, you specify the prior conversational turns with the `messages` parameter, and the model then generates the next `Message` in the conversation. Consecutive `user` or `assistant` turns in your request will be combined into a single turn. Each input message must be an object with a `role` and `content`. You can specify a single `user`-role message, or you can include multiple `user` and `assistant` messages. If the final message uses the `assistant` role, the response content will continue immediately from the content in that message. This can be used to constrain part of the model's response. Example with a single `user` message: ```json [{ "role": "user", "content": "Hello, Claude" }] ``` Example with multiple conversational turns: ```json [ { "role": "user", "content": "Hello there." }, { "role": "assistant", "content": "Hi, I'm Claude. How can I help you?" }, { "role": "user", "content": "Can you explain LLMs in plain English?" } ] ``` Example with a partially-filled response from Claude: ```json [ { "role": "user", "content": "What's the Greek name for Sun? (A) Sol (B) Helios (C) Sun" }, { "role": "assistant", "content": "The best answer is (" } ] ``` Each input message `content` may be either a single `string` or an array of content blocks, where each block has a specific `type`. Using a `string` for `content` is shorthand for an array of one content block of type `"text"`. The following input messages are equivalent: ```json { "role": "user", "content": "Hello, Claude" } ``` ```json { "role": "user", "content": [{ "type": "text", "text": "Hello, Claude" }] } ``` See [input examples](https://docs.claude.com/en/api/messages-examples). Note that if you want to include a [system prompt](https://docs.claude.com/en/docs/system-prompts), you can use the top-level `system` parameter — there is no `"system"` role for input messages in the Messages API. There is a limit of 100,000 messages in a single request. model: The model that will complete your prompt.\n\nSee [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. stream: Whether to incrementally stream the response using server-sent events. See [streaming](https://docs.claude.com/en/api/messages-streaming) for details. container: Container identifier for reuse across requests. context_management: Context management configuration. This allows you to control how Claude manages context across multiple requests, such as whether to clear function results or not. mcp_servers: MCP servers to be utilized in this request metadata: An object describing metadata about the request. output_config: Configuration options for the model's output. Controls aspects like how much effort the model puts into its response. output_format: A schema to specify Claude's output format in responses. service_tier: Determines whether to use priority capacity (if available) or standard capacity for this request. Anthropic offers different levels of service for your API requests. See [service-tiers](https://docs.claude.com/en/api/service-tiers) for details. stop_sequences: Custom text sequences that will cause the model to stop generating. Our models will normally stop when they have naturally completed their turn, which will result in a response `stop_reason` of `"end_turn"`. If you want the model to stop generating when it encounters custom strings of text, you can use the `stop_sequences` parameter. If the model encounters one of the custom sequences, the response `stop_reason` value will be `"stop_sequence"` and the response `stop_sequence` value will contain the matched stop sequence. system: System prompt. A system prompt is a way of providing context and instructions to Claude, such as specifying a particular goal or role. See our [guide to system prompts](https://docs.claude.com/en/docs/system-prompts). temperature: Amount of randomness injected into the response. Defaults to `1.0`. Ranges from `0.0` to `1.0`. Use `temperature` closer to `0.0` for analytical / multiple choice, and closer to `1.0` for creative and generative tasks. Note that even with `temperature` of `0.0`, the results will not be fully deterministic. thinking: Configuration for enabling Claude's extended thinking. When enabled, responses include `thinking` content blocks showing Claude's thinking process before the final answer. Requires a minimum budget of 1,024 tokens and counts towards your `max_tokens` limit. See [extended thinking](https://docs.claude.com/en/docs/build-with-claude/extended-thinking) for details. tool_choice: How the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all. tools: Definitions of tools that the model may use. If you include `tools` in your API request, the model may return `tool_use` content blocks that represent the model's use of those tools. You can then run those tools using the tool input generated by the model and then optionally return results back to the model using `tool_result` content blocks. There are two types of tools: **client tools** and **server tools**. The behavior described below applies to client tools. For [server tools](https://docs.claude.com/en/docs/agents-and-tools/tool-use/overview#server-tools), see their individual documentation as each has its own behavior (e.g., the [web search tool](https://docs.claude.com/en/docs/agents-and-tools/tool-use/web-search-tool)). Each tool definition includes: - `name`: Name of the tool. - `description`: Optional, but strongly-recommended description of the tool. - `input_schema`: [JSON schema](https://json-schema.org/draft/2020-12) for the tool `input` shape that the model will produce in `tool_use` output content blocks. For example, if you defined `tools` as: ```json [ { "name": "get_stock_price", "description": "Get the current stock price for a given ticker symbol.", "input_schema": { "type": "object", "properties": { "ticker": { "type": "string", "description": "The stock ticker symbol, e.g. AAPL for Apple Inc." } }, "required": ["ticker"] } } ] ``` And then asked the model "What's the S&P 500 at today?", the model might produce `tool_use` content blocks in the response like this: ```json [ { "type": "tool_use", "id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "name": "get_stock_price", "input": { "ticker": "^GSPC" } } ] ``` You might then run your `get_stock_price` tool with `{"ticker": "^GSPC"}` as an input, and return the following back to the model in a subsequent `user` message: ```json [ { "type": "tool_result", "tool_use_id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "content": "259.75 USD" } ] ``` Tools can be used for workflows that include running client-side tools and functions, or more generally whenever you want the model to produce a particular JSON structure of output. See our [guide](https://docs.claude.com/en/docs/tool-use) for more details. top_k: Only sample from the top K options for each subsequent token. Used to remove "long tail" low probability responses. [Learn more technical details here](https://towardsdatascience.com/how-to-sample-from-language-models-682bceb97277). Recommended for advanced use cases only. You usually only need to use `temperature`. top_p: Use nucleus sampling. In nucleus sampling, we compute the cumulative distribution over all the options for each subsequent token in decreasing probability order and cut it off once it reaches a particular probability specified by `top_p`. You should either alter `temperature` or `top_p`, but not both. Recommended for advanced use cases only. You usually only need to use `temperature`. betas: Optional header to specify the beta version(s) you want to use. extra_headers: Send extra headers extra_query: Add additional query parameters to the request extra_body: Add additional JSON properties to the request timeout: Override the client-level default timeout for this request, in seconds """ ... @overload def create( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, stream: bool, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> BetaMessage | Stream[BetaRawMessageStreamEvent]: """ Send a structured list of input messages with text and/or image content, and the model will generate the next message in the conversation. The Messages API can be used for either single queries or stateless multi-turn conversations. Learn more about the Messages API in our [user guide](https://docs.claude.com/en/docs/initial-setup) Args: max_tokens: The maximum number of tokens to generate before stopping. Note that our models may stop _before_ reaching this maximum. This parameter only specifies the absolute maximum number of tokens to generate. Different models have different maximum values for this parameter. See [models](https://docs.claude.com/en/docs/models-overview) for details. messages: Input messages. Our models are trained to operate on alternating `user` and `assistant` conversational turns. When creating a new `Message`, you specify the prior conversational turns with the `messages` parameter, and the model then generates the next `Message` in the conversation. Consecutive `user` or `assistant` turns in your request will be combined into a single turn. Each input message must be an object with a `role` and `content`. You can specify a single `user`-role message, or you can include multiple `user` and `assistant` messages. If the final message uses the `assistant` role, the response content will continue immediately from the content in that message. This can be used to constrain part of the model's response. Example with a single `user` message: ```json [{ "role": "user", "content": "Hello, Claude" }] ``` Example with multiple conversational turns: ```json [ { "role": "user", "content": "Hello there." }, { "role": "assistant", "content": "Hi, I'm Claude. How can I help you?" }, { "role": "user", "content": "Can you explain LLMs in plain English?" } ] ``` Example with a partially-filled response from Claude: ```json [ { "role": "user", "content": "What's the Greek name for Sun? (A) Sol (B) Helios (C) Sun" }, { "role": "assistant", "content": "The best answer is (" } ] ``` Each input message `content` may be either a single `string` or an array of content blocks, where each block has a specific `type`. Using a `string` for `content` is shorthand for an array of one content block of type `"text"`. The following input messages are equivalent: ```json { "role": "user", "content": "Hello, Claude" } ``` ```json { "role": "user", "content": [{ "type": "text", "text": "Hello, Claude" }] } ``` See [input examples](https://docs.claude.com/en/api/messages-examples). Note that if you want to include a [system prompt](https://docs.claude.com/en/docs/system-prompts), you can use the top-level `system` parameter — there is no `"system"` role for input messages in the Messages API. There is a limit of 100,000 messages in a single request. model: The model that will complete your prompt.\n\nSee [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. stream: Whether to incrementally stream the response using server-sent events. See [streaming](https://docs.claude.com/en/api/messages-streaming) for details. container: Container identifier for reuse across requests. context_management: Context management configuration. This allows you to control how Claude manages context across multiple requests, such as whether to clear function results or not. mcp_servers: MCP servers to be utilized in this request metadata: An object describing metadata about the request. output_config: Configuration options for the model's output. Controls aspects like how much effort the model puts into its response. output_format: A schema to specify Claude's output format in responses. service_tier: Determines whether to use priority capacity (if available) or standard capacity for this request. Anthropic offers different levels of service for your API requests. See [service-tiers](https://docs.claude.com/en/api/service-tiers) for details. stop_sequences: Custom text sequences that will cause the model to stop generating. Our models will normally stop when they have naturally completed their turn, which will result in a response `stop_reason` of `"end_turn"`. If you want the model to stop generating when it encounters custom strings of text, you can use the `stop_sequences` parameter. If the model encounters one of the custom sequences, the response `stop_reason` value will be `"stop_sequence"` and the response `stop_sequence` value will contain the matched stop sequence. system: System prompt. A system prompt is a way of providing context and instructions to Claude, such as specifying a particular goal or role. See our [guide to system prompts](https://docs.claude.com/en/docs/system-prompts). temperature: Amount of randomness injected into the response. Defaults to `1.0`. Ranges from `0.0` to `1.0`. Use `temperature` closer to `0.0` for analytical / multiple choice, and closer to `1.0` for creative and generative tasks. Note that even with `temperature` of `0.0`, the results will not be fully deterministic. thinking: Configuration for enabling Claude's extended thinking. When enabled, responses include `thinking` content blocks showing Claude's thinking process before the final answer. Requires a minimum budget of 1,024 tokens and counts towards your `max_tokens` limit. See [extended thinking](https://docs.claude.com/en/docs/build-with-claude/extended-thinking) for details. tool_choice: How the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all. tools: Definitions of tools that the model may use. If you include `tools` in your API request, the model may return `tool_use` content blocks that represent the model's use of those tools. You can then run those tools using the tool input generated by the model and then optionally return results back to the model using `tool_result` content blocks. There are two types of tools: **client tools** and **server tools**. The behavior described below applies to client tools. For [server tools](https://docs.claude.com/en/docs/agents-and-tools/tool-use/overview#server-tools), see their individual documentation as each has its own behavior (e.g., the [web search tool](https://docs.claude.com/en/docs/agents-and-tools/tool-use/web-search-tool)). Each tool definition includes: - `name`: Name of the tool. - `description`: Optional, but strongly-recommended description of the tool. - `input_schema`: [JSON schema](https://json-schema.org/draft/2020-12) for the tool `input` shape that the model will produce in `tool_use` output content blocks. For example, if you defined `tools` as: ```json [ { "name": "get_stock_price", "description": "Get the current stock price for a given ticker symbol.", "input_schema": { "type": "object", "properties": { "ticker": { "type": "string", "description": "The stock ticker symbol, e.g. AAPL for Apple Inc." } }, "required": ["ticker"] } } ] ``` And then asked the model "What's the S&P 500 at today?", the model might produce `tool_use` content blocks in the response like this: ```json [ { "type": "tool_use", "id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "name": "get_stock_price", "input": { "ticker": "^GSPC" } } ] ``` You might then run your `get_stock_price` tool with `{"ticker": "^GSPC"}` as an input, and return the following back to the model in a subsequent `user` message: ```json [ { "type": "tool_result", "tool_use_id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "content": "259.75 USD" } ] ``` Tools can be used for workflows that include running client-side tools and functions, or more generally whenever you want the model to produce a particular JSON structure of output. See our [guide](https://docs.claude.com/en/docs/tool-use) for more details. top_k: Only sample from the top K options for each subsequent token. Used to remove "long tail" low probability responses. [Learn more technical details here](https://towardsdatascience.com/how-to-sample-from-language-models-682bceb97277). Recommended for advanced use cases only. You usually only need to use `temperature`. top_p: Use nucleus sampling. In nucleus sampling, we compute the cumulative distribution over all the options for each subsequent token in decreasing probability order and cut it off once it reaches a particular probability specified by `top_p`. You should either alter `temperature` or `top_p`, but not both. Recommended for advanced use cases only. You usually only need to use `temperature`. betas: Optional header to specify the beta version(s) you want to use. extra_headers: Send extra headers extra_query: Add additional query parameters to the request extra_body: Add additional JSON properties to the request timeout: Override the client-level default timeout for this request, in seconds """ ... @required_args(["max_tokens", "messages", "model"], ["max_tokens", "messages", "model", "stream"]) def create( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: Literal[False] | Literal[True] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> BetaMessage | Stream[BetaRawMessageStreamEvent]: validate_output_format(output_format) if not stream and not is_given(timeout) and self._client.timeout == DEFAULT_TIMEOUT: timeout = self._client._calculate_nonstreaming_timeout( max_tokens, MODEL_NONSTREAMING_TOKENS.get(model, None) ) if model in DEPRECATED_MODELS: warnings.warn( f"The model '{model}' is deprecated and will reach end-of-life on {DEPRECATED_MODELS[model]}.\nPlease migrate to a newer model. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.", DeprecationWarning, stacklevel=3, ) extra_headers = { **strip_not_given({"anthropic-beta": ",".join(str(e) for e in betas) if is_given(betas) else not_given}), **(extra_headers or {}), } return self._post( "/v1/messages?beta=true", body=maybe_transform( { "max_tokens": max_tokens, "messages": messages, "model": model, "container": container, "context_management": context_management, "mcp_servers": mcp_servers, "metadata": metadata, "output_config": output_config, "output_format": output_format, "service_tier": service_tier, "stop_sequences": stop_sequences, "stream": stream, "system": system, "temperature": temperature, "thinking": thinking, "tool_choice": tool_choice, "tools": tools, "top_k": top_k, "top_p": top_p, }, message_create_params.MessageCreateParamsStreaming if stream else message_create_params.MessageCreateParamsNonStreaming, ), options=make_request_options( extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout ), cast_to=BetaMessage, stream=stream or False, stream_cls=Stream[BetaRawMessageStreamEvent], ) def parse( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: Literal[False] | Literal[True] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> ParsedBetaMessage[ResponseFormatT]: if not stream and not is_given(timeout) and self._client.timeout == DEFAULT_TIMEOUT: timeout = self._client._calculate_nonstreaming_timeout( max_tokens, MODEL_NONSTREAMING_TOKENS.get(model, None) ) if model in DEPRECATED_MODELS: warnings.warn( f"The model '{model}' is deprecated and will reach end-of-life on {DEPRECATED_MODELS[model]}.\nPlease migrate to a newer model. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.", DeprecationWarning, stacklevel=3, ) betas = [beta for beta in betas] if is_given(betas) else [] if "structured-outputs-2025-11-13" not in betas: # Ensure structured outputs beta is included for parse method betas.append("structured-outputs-2025-11-13") extra_headers = { "X-Stainless-Helper": "beta.messages.parse", **strip_not_given({"anthropic-beta": ",".join(str(e) for e in betas) if is_given(betas) else NOT_GIVEN}), **(extra_headers or {}), } transformed_output_format: Optional[message_create_params.OutputFormat] | NotGiven = NOT_GIVEN if is_given(output_format) and output_format is not None: adapted_type: TypeAdapter[ResponseFormatT] = TypeAdapter(output_format) try: schema = adapted_type.json_schema() transformed_output_format = message_create_params.OutputFormat( schema=transform_schema(schema), type="json_schema" ) except pydantic.errors.PydanticSchemaGenerationError as e: raise TypeError( ( "Could not generate JSON schema for the given `output_format` type. " "Use a type that works with `pydanitc.TypeAdapter`" ) ) from e def parser(response: BetaMessage) -> ParsedBetaMessage[ResponseFormatT]: return parse_response( response=response, output_format=cast( ResponseFormatT, output_format if is_given(output_format) and output_format is not None else NOT_GIVEN, ), ) return self._post( "/v1/messages?beta=true", body=maybe_transform( { "max_tokens": max_tokens, "messages": messages, "model": model, "container": container, "context_management": context_management, "mcp_servers": mcp_servers, "metadata": metadata, "output_config": output_config, "output_format": transformed_output_format, "service_tier": service_tier, "stop_sequences": stop_sequences, "stream": stream, "system": system, "temperature": temperature, "thinking": thinking, "tool_choice": tool_choice, "tools": tools, "top_k": top_k, "top_p": top_p, }, message_create_params.MessageCreateParamsNonStreaming, ), options=make_request_options( extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout, post_parser=parser, ), cast_to=cast(Type[ParsedBetaMessage[ResponseFormatT]], BetaMessage), stream=False, ) @overload def tool_runner( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, tools: Iterable[BetaRunnableTool], compaction_control: CompactionControl | Omit = omit, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, max_iterations: int | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: Literal[False] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaToolRunner[ResponseFormatT]: ... @overload def tool_runner( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, tools: Iterable[BetaRunnableTool], compaction_control: CompactionControl | Omit = omit, stream: Literal[True], max_iterations: int | Omit = omit, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaStreamingToolRunner[ResponseFormatT]: ... @overload def tool_runner( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, tools: Iterable[BetaRunnableTool], compaction_control: CompactionControl | Omit = omit, stream: bool, max_iterations: int | Omit = omit, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaStreamingToolRunner[ResponseFormatT] | BetaToolRunner[ResponseFormatT]: ... def tool_runner( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, tools: Iterable[BetaRunnableTool], compaction_control: CompactionControl | Omit = omit, max_iterations: int | Omit = omit, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: bool | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaStreamingToolRunner[ResponseFormatT] | BetaToolRunner[ResponseFormatT]: """Create a Message stream""" if model in DEPRECATED_MODELS: warnings.warn( f"The model '{model}' is deprecated and will reach end-of-life on {DEPRECATED_MODELS[model]}.\nPlease migrate to a newer model. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.", DeprecationWarning, stacklevel=3, ) extra_headers = { "X-Stainless-Helper": "BetaToolRunner", **strip_not_given({"anthropic-beta": ",".join(str(e) for e in betas) if is_given(betas) else NOT_GIVEN}), **(extra_headers or {}), } params = cast( message_create_params.ParseMessageCreateParamsBase[ResponseFormatT], { "max_tokens": max_tokens, "messages": messages, "model": model, "container": container, "context_management": context_management, "mcp_servers": mcp_servers, "metadata": metadata, "output_config": output_config, "output_format": output_format, "service_tier": service_tier, "stop_sequences": stop_sequences, "system": system, "temperature": temperature, "thinking": thinking, "tool_choice": tool_choice, "tools": [tool.to_dict() for tool in tools], "top_k": top_k, "top_p": top_p, }, ) if stream: return BetaStreamingToolRunner[ResponseFormatT]( tools=tools, params=params, options={ "extra_headers": extra_headers, "extra_query": extra_query, "extra_body": extra_body, "timeout": timeout, }, client=cast("Anthropic", self._client), max_iterations=max_iterations if is_given(max_iterations) else None, compaction_control=compaction_control if is_given(compaction_control) else None, ) return BetaToolRunner[ResponseFormatT]( tools=tools, params=params, options={ "extra_headers": extra_headers, "extra_query": extra_query, "extra_body": extra_body, "timeout": timeout, }, client=cast("Anthropic", self._client), max_iterations=max_iterations if is_given(max_iterations) else None, compaction_control=compaction_control if is_given(compaction_control) else None, ) def stream( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaMessageStreamManager[ResponseFormatT]: if model in DEPRECATED_MODELS: warnings.warn( f"The model '{model}' is deprecated and will reach end-of-life on {DEPRECATED_MODELS[model]}.\nPlease migrate to a newer model. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.", DeprecationWarning, stacklevel=3, ) """Create a Message stream""" extra_headers = { "X-Stainless-Helper-Method": "stream", "X-Stainless-Stream-Helper": "beta.messages", **strip_not_given({"anthropic-beta": ",".join(str(e) for e in betas) if is_given(betas) else NOT_GIVEN}), **(extra_headers or {}), } transformed_output_format: Optional[message_create_params.OutputFormat] | NotGiven = NOT_GIVEN if is_given(output_format) and output_format is not None: adapted_type: TypeAdapter[ResponseFormatT] = TypeAdapter(output_format) try: schema = adapted_type.json_schema() transformed_output_format = message_create_params.OutputFormat( schema=transform_schema(schema), type="json_schema" ) except pydantic.errors.PydanticSchemaGenerationError as e: raise TypeError( ( "Could not generate JSON schema for the given `output_format` type. " "Use a type that works with `pydanitc.TypeAdapter`" ) ) from e make_request = partial( self._post, "/v1/messages?beta=true", body=maybe_transform( { "max_tokens": max_tokens, "messages": messages, "model": model, "metadata": metadata, "output_config": output_config, "output_format": transformed_output_format, "container": container, "context_management": context_management, "mcp_servers": mcp_servers, "service_tier": service_tier, "stop_sequences": stop_sequences, "system": system, "temperature": temperature, "thinking": thinking, "top_k": top_k, "top_p": top_p, "tools": tools, "tool_choice": tool_choice, "stream": True, }, message_create_params.MessageCreateParams, ), options=make_request_options( extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout ), cast_to=BetaMessage, stream=True, stream_cls=Stream[BetaRawMessageStreamEvent], ) return BetaMessageStreamManager(make_request, output_format=cast(ResponseFormatT, output_format)) def count_tokens( self, *, messages: Iterable[BetaMessageParam], model: ModelParam, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[message_count_tokens_params.Tool] | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> BetaMessageTokensCount: """ Count the number of tokens in a Message. The Token Count API can be used to count the number of tokens in a Message, including tools, images, and documents, without creating it. Learn more about token counting in our [user guide](https://docs.claude.com/en/docs/build-with-claude/token-counting) Args: messages: Input messages. Our models are trained to operate on alternating `user` and `assistant` conversational turns. When creating a new `Message`, you specify the prior conversational turns with the `messages` parameter, and the model then generates the next `Message` in the conversation. Consecutive `user` or `assistant` turns in your request will be combined into a single turn. Each input message must be an object with a `role` and `content`. You can specify a single `user`-role message, or you can include multiple `user` and `assistant` messages. If the final message uses the `assistant` role, the response content will continue immediately from the content in that message. This can be used to constrain part of the model's response. Example with a single `user` message: ```json [{ "role": "user", "content": "Hello, Claude" }] ``` Example with multiple conversational turns: ```json [ { "role": "user", "content": "Hello there." }, { "role": "assistant", "content": "Hi, I'm Claude. How can I help you?" }, { "role": "user", "content": "Can you explain LLMs in plain English?" } ] ``` Example with a partially-filled response from Claude: ```json [ { "role": "user", "content": "What's the Greek name for Sun? (A) Sol (B) Helios (C) Sun" }, { "role": "assistant", "content": "The best answer is (" } ] ``` Each input message `content` may be either a single `string` or an array of content blocks, where each block has a specific `type`. Using a `string` for `content` is shorthand for an array of one content block of type `"text"`. The following input messages are equivalent: ```json { "role": "user", "content": "Hello, Claude" } ``` ```json { "role": "user", "content": [{ "type": "text", "text": "Hello, Claude" }] } ``` See [input examples](https://docs.claude.com/en/api/messages-examples). Note that if you want to include a [system prompt](https://docs.claude.com/en/docs/system-prompts), you can use the top-level `system` parameter — there is no `"system"` role for input messages in the Messages API. There is a limit of 100,000 messages in a single request. model: The model that will complete your prompt.\n\nSee [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. context_management: Context management configuration. This allows you to control how Claude manages context across multiple requests, such as whether to clear function results or not. mcp_servers: MCP servers to be utilized in this request output_config: Configuration options for the model's output. Controls aspects like how much effort the model puts into its response. output_format: A schema to specify Claude's output format in responses. system: System prompt. A system prompt is a way of providing context and instructions to Claude, such as specifying a particular goal or role. See our [guide to system prompts](https://docs.claude.com/en/docs/system-prompts). thinking: Configuration for enabling Claude's extended thinking. When enabled, responses include `thinking` content blocks showing Claude's thinking process before the final answer. Requires a minimum budget of 1,024 tokens and counts towards your `max_tokens` limit. See [extended thinking](https://docs.claude.com/en/docs/build-with-claude/extended-thinking) for details. tool_choice: How the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all. tools: Definitions of tools that the model may use. If you include `tools` in your API request, the model may return `tool_use` content blocks that represent the model's use of those tools. You can then run those tools using the tool input generated by the model and then optionally return results back to the model using `tool_result` content blocks. There are two types of tools: **client tools** and **server tools**. The behavior described below applies to client tools. For [server tools](https://docs.claude.com/en/docs/agents-and-tools/tool-use/overview#server-tools), see their individual documentation as each has its own behavior (e.g., the [web search tool](https://docs.claude.com/en/docs/agents-and-tools/tool-use/web-search-tool)). Each tool definition includes: - `name`: Name of the tool. - `description`: Optional, but strongly-recommended description of the tool. - `input_schema`: [JSON schema](https://json-schema.org/draft/2020-12) for the tool `input` shape that the model will produce in `tool_use` output content blocks. For example, if you defined `tools` as: ```json [ { "name": "get_stock_price", "description": "Get the current stock price for a given ticker symbol.", "input_schema": { "type": "object", "properties": { "ticker": { "type": "string", "description": "The stock ticker symbol, e.g. AAPL for Apple Inc." } }, "required": ["ticker"] } } ] ``` And then asked the model "What's the S&P 500 at today?", the model might produce `tool_use` content blocks in the response like this: ```json [ { "type": "tool_use", "id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "name": "get_stock_price", "input": { "ticker": "^GSPC" } } ] ``` You might then run your `get_stock_price` tool with `{"ticker": "^GSPC"}` as an input, and return the following back to the model in a subsequent `user` message: ```json [ { "type": "tool_result", "tool_use_id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "content": "259.75 USD" } ] ``` Tools can be used for workflows that include running client-side tools and functions, or more generally whenever you want the model to produce a particular JSON structure of output. See our [guide](https://docs.claude.com/en/docs/tool-use) for more details. betas: Optional header to specify the beta version(s) you want to use. extra_headers: Send extra headers extra_query: Add additional query parameters to the request extra_body: Add additional JSON properties to the request timeout: Override the client-level default timeout for this request, in seconds """ extra_headers = { **strip_not_given( { "anthropic-beta": ",".join(chain((str(e) for e in betas), ["token-counting-2024-11-01"])) if is_given(betas) else not_given } ), **(extra_headers or {}), } extra_headers = {"anthropic-beta": "token-counting-2024-11-01", **(extra_headers or {})} return self._post( "/v1/messages/count_tokens?beta=true", body=maybe_transform( { "messages": messages, "model": model, "context_management": context_management, "mcp_servers": mcp_servers, "output_config": output_config, "output_format": output_format, "system": system, "thinking": thinking, "tool_choice": tool_choice, "tools": tools, }, message_count_tokens_params.MessageCountTokensParams, ), options=make_request_options( extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout ), cast_to=BetaMessageTokensCount, ) class AsyncMessages(AsyncAPIResource): @cached_property def batches(self) -> AsyncBatches: return AsyncBatches(self._client) @cached_property def with_raw_response(self) -> AsyncMessagesWithRawResponse: """ This property can be used as a prefix for any HTTP method call to return the raw response object instead of the parsed content. For more information, see https://www.github.com/anthropics/anthropic-sdk-python#accessing-raw-response-data-eg-headers """ return AsyncMessagesWithRawResponse(self) @cached_property def with_streaming_response(self) -> AsyncMessagesWithStreamingResponse: """ An alternative to `.with_raw_response` that doesn't eagerly read the response body. For more information, see https://www.github.com/anthropics/anthropic-sdk-python#with_streaming_response """ return AsyncMessagesWithStreamingResponse(self) @overload async def create( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: Literal[False] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> BetaMessage: """ Send a structured list of input messages with text and/or image content, and the model will generate the next message in the conversation. The Messages API can be used for either single queries or stateless multi-turn conversations. Learn more about the Messages API in our [user guide](https://docs.claude.com/en/docs/initial-setup) Args: max_tokens: The maximum number of tokens to generate before stopping. Note that our models may stop _before_ reaching this maximum. This parameter only specifies the absolute maximum number of tokens to generate. Different models have different maximum values for this parameter. See [models](https://docs.claude.com/en/docs/models-overview) for details. messages: Input messages. Our models are trained to operate on alternating `user` and `assistant` conversational turns. When creating a new `Message`, you specify the prior conversational turns with the `messages` parameter, and the model then generates the next `Message` in the conversation. Consecutive `user` or `assistant` turns in your request will be combined into a single turn. Each input message must be an object with a `role` and `content`. You can specify a single `user`-role message, or you can include multiple `user` and `assistant` messages. If the final message uses the `assistant` role, the response content will continue immediately from the content in that message. This can be used to constrain part of the model's response. Example with a single `user` message: ```json [{ "role": "user", "content": "Hello, Claude" }] ``` Example with multiple conversational turns: ```json [ { "role": "user", "content": "Hello there." }, { "role": "assistant", "content": "Hi, I'm Claude. How can I help you?" }, { "role": "user", "content": "Can you explain LLMs in plain English?" } ] ``` Example with a partially-filled response from Claude: ```json [ { "role": "user", "content": "What's the Greek name for Sun? (A) Sol (B) Helios (C) Sun" }, { "role": "assistant", "content": "The best answer is (" } ] ``` Each input message `content` may be either a single `string` or an array of content blocks, where each block has a specific `type`. Using a `string` for `content` is shorthand for an array of one content block of type `"text"`. The following input messages are equivalent: ```json { "role": "user", "content": "Hello, Claude" } ``` ```json { "role": "user", "content": [{ "type": "text", "text": "Hello, Claude" }] } ``` See [input examples](https://docs.claude.com/en/api/messages-examples). Note that if you want to include a [system prompt](https://docs.claude.com/en/docs/system-prompts), you can use the top-level `system` parameter — there is no `"system"` role for input messages in the Messages API. There is a limit of 100,000 messages in a single request. model: The model that will complete your prompt.\n\nSee [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. container: Container identifier for reuse across requests. context_management: Context management configuration. This allows you to control how Claude manages context across multiple requests, such as whether to clear function results or not. mcp_servers: MCP servers to be utilized in this request metadata: An object describing metadata about the request. output_config: Configuration options for the model's output. Controls aspects like how much effort the model puts into its response. output_format: A schema to specify Claude's output format in responses. service_tier: Determines whether to use priority capacity (if available) or standard capacity for this request. Anthropic offers different levels of service for your API requests. See [service-tiers](https://docs.claude.com/en/api/service-tiers) for details. stop_sequences: Custom text sequences that will cause the model to stop generating. Our models will normally stop when they have naturally completed their turn, which will result in a response `stop_reason` of `"end_turn"`. If you want the model to stop generating when it encounters custom strings of text, you can use the `stop_sequences` parameter. If the model encounters one of the custom sequences, the response `stop_reason` value will be `"stop_sequence"` and the response `stop_sequence` value will contain the matched stop sequence. stream: Whether to incrementally stream the response using server-sent events. See [streaming](https://docs.claude.com/en/api/messages-streaming) for details. system: System prompt. A system prompt is a way of providing context and instructions to Claude, such as specifying a particular goal or role. See our [guide to system prompts](https://docs.claude.com/en/docs/system-prompts). temperature: Amount of randomness injected into the response. Defaults to `1.0`. Ranges from `0.0` to `1.0`. Use `temperature` closer to `0.0` for analytical / multiple choice, and closer to `1.0` for creative and generative tasks. Note that even with `temperature` of `0.0`, the results will not be fully deterministic. thinking: Configuration for enabling Claude's extended thinking. When enabled, responses include `thinking` content blocks showing Claude's thinking process before the final answer. Requires a minimum budget of 1,024 tokens and counts towards your `max_tokens` limit. See [extended thinking](https://docs.claude.com/en/docs/build-with-claude/extended-thinking) for details. tool_choice: How the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all. tools: Definitions of tools that the model may use. If you include `tools` in your API request, the model may return `tool_use` content blocks that represent the model's use of those tools. You can then run those tools using the tool input generated by the model and then optionally return results back to the model using `tool_result` content blocks. There are two types of tools: **client tools** and **server tools**. The behavior described below applies to client tools. For [server tools](https://docs.claude.com/en/docs/agents-and-tools/tool-use/overview#server-tools), see their individual documentation as each has its own behavior (e.g., the [web search tool](https://docs.claude.com/en/docs/agents-and-tools/tool-use/web-search-tool)). Each tool definition includes: - `name`: Name of the tool. - `description`: Optional, but strongly-recommended description of the tool. - `input_schema`: [JSON schema](https://json-schema.org/draft/2020-12) for the tool `input` shape that the model will produce in `tool_use` output content blocks. For example, if you defined `tools` as: ```json [ { "name": "get_stock_price", "description": "Get the current stock price for a given ticker symbol.", "input_schema": { "type": "object", "properties": { "ticker": { "type": "string", "description": "The stock ticker symbol, e.g. AAPL for Apple Inc." } }, "required": ["ticker"] } } ] ``` And then asked the model "What's the S&P 500 at today?", the model might produce `tool_use` content blocks in the response like this: ```json [ { "type": "tool_use", "id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "name": "get_stock_price", "input": { "ticker": "^GSPC" } } ] ``` You might then run your `get_stock_price` tool with `{"ticker": "^GSPC"}` as an input, and return the following back to the model in a subsequent `user` message: ```json [ { "type": "tool_result", "tool_use_id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "content": "259.75 USD" } ] ``` Tools can be used for workflows that include running client-side tools and functions, or more generally whenever you want the model to produce a particular JSON structure of output. See our [guide](https://docs.claude.com/en/docs/tool-use) for more details. top_k: Only sample from the top K options for each subsequent token. Used to remove "long tail" low probability responses. [Learn more technical details here](https://towardsdatascience.com/how-to-sample-from-language-models-682bceb97277). Recommended for advanced use cases only. You usually only need to use `temperature`. top_p: Use nucleus sampling. In nucleus sampling, we compute the cumulative distribution over all the options for each subsequent token in decreasing probability order and cut it off once it reaches a particular probability specified by `top_p`. You should either alter `temperature` or `top_p`, but not both. Recommended for advanced use cases only. You usually only need to use `temperature`. betas: Optional header to specify the beta version(s) you want to use. extra_headers: Send extra headers extra_query: Add additional query parameters to the request extra_body: Add additional JSON properties to the request timeout: Override the client-level default timeout for this request, in seconds """ ... @overload async def create( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, stream: Literal[True], container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> AsyncStream[BetaRawMessageStreamEvent]: """ Send a structured list of input messages with text and/or image content, and the model will generate the next message in the conversation. The Messages API can be used for either single queries or stateless multi-turn conversations. Learn more about the Messages API in our [user guide](https://docs.claude.com/en/docs/initial-setup) Args: max_tokens: The maximum number of tokens to generate before stopping. Note that our models may stop _before_ reaching this maximum. This parameter only specifies the absolute maximum number of tokens to generate. Different models have different maximum values for this parameter. See [models](https://docs.claude.com/en/docs/models-overview) for details. messages: Input messages. Our models are trained to operate on alternating `user` and `assistant` conversational turns. When creating a new `Message`, you specify the prior conversational turns with the `messages` parameter, and the model then generates the next `Message` in the conversation. Consecutive `user` or `assistant` turns in your request will be combined into a single turn. Each input message must be an object with a `role` and `content`. You can specify a single `user`-role message, or you can include multiple `user` and `assistant` messages. If the final message uses the `assistant` role, the response content will continue immediately from the content in that message. This can be used to constrain part of the model's response. Example with a single `user` message: ```json [{ "role": "user", "content": "Hello, Claude" }] ``` Example with multiple conversational turns: ```json [ { "role": "user", "content": "Hello there." }, { "role": "assistant", "content": "Hi, I'm Claude. How can I help you?" }, { "role": "user", "content": "Can you explain LLMs in plain English?" } ] ``` Example with a partially-filled response from Claude: ```json [ { "role": "user", "content": "What's the Greek name for Sun? (A) Sol (B) Helios (C) Sun" }, { "role": "assistant", "content": "The best answer is (" } ] ``` Each input message `content` may be either a single `string` or an array of content blocks, where each block has a specific `type`. Using a `string` for `content` is shorthand for an array of one content block of type `"text"`. The following input messages are equivalent: ```json { "role": "user", "content": "Hello, Claude" } ``` ```json { "role": "user", "content": [{ "type": "text", "text": "Hello, Claude" }] } ``` See [input examples](https://docs.claude.com/en/api/messages-examples). Note that if you want to include a [system prompt](https://docs.claude.com/en/docs/system-prompts), you can use the top-level `system` parameter — there is no `"system"` role for input messages in the Messages API. There is a limit of 100,000 messages in a single request. model: The model that will complete your prompt.\n\nSee [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. stream: Whether to incrementally stream the response using server-sent events. See [streaming](https://docs.claude.com/en/api/messages-streaming) for details. container: Container identifier for reuse across requests. context_management: Context management configuration. This allows you to control how Claude manages context across multiple requests, such as whether to clear function results or not. mcp_servers: MCP servers to be utilized in this request metadata: An object describing metadata about the request. output_config: Configuration options for the model's output. Controls aspects like how much effort the model puts into its response. output_format: A schema to specify Claude's output format in responses. service_tier: Determines whether to use priority capacity (if available) or standard capacity for this request. Anthropic offers different levels of service for your API requests. See [service-tiers](https://docs.claude.com/en/api/service-tiers) for details. stop_sequences: Custom text sequences that will cause the model to stop generating. Our models will normally stop when they have naturally completed their turn, which will result in a response `stop_reason` of `"end_turn"`. If you want the model to stop generating when it encounters custom strings of text, you can use the `stop_sequences` parameter. If the model encounters one of the custom sequences, the response `stop_reason` value will be `"stop_sequence"` and the response `stop_sequence` value will contain the matched stop sequence. system: System prompt. A system prompt is a way of providing context and instructions to Claude, such as specifying a particular goal or role. See our [guide to system prompts](https://docs.claude.com/en/docs/system-prompts). temperature: Amount of randomness injected into the response. Defaults to `1.0`. Ranges from `0.0` to `1.0`. Use `temperature` closer to `0.0` for analytical / multiple choice, and closer to `1.0` for creative and generative tasks. Note that even with `temperature` of `0.0`, the results will not be fully deterministic. thinking: Configuration for enabling Claude's extended thinking. When enabled, responses include `thinking` content blocks showing Claude's thinking process before the final answer. Requires a minimum budget of 1,024 tokens and counts towards your `max_tokens` limit. See [extended thinking](https://docs.claude.com/en/docs/build-with-claude/extended-thinking) for details. tool_choice: How the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all. tools: Definitions of tools that the model may use. If you include `tools` in your API request, the model may return `tool_use` content blocks that represent the model's use of those tools. You can then run those tools using the tool input generated by the model and then optionally return results back to the model using `tool_result` content blocks. There are two types of tools: **client tools** and **server tools**. The behavior described below applies to client tools. For [server tools](https://docs.claude.com/en/docs/agents-and-tools/tool-use/overview#server-tools), see their individual documentation as each has its own behavior (e.g., the [web search tool](https://docs.claude.com/en/docs/agents-and-tools/tool-use/web-search-tool)). Each tool definition includes: - `name`: Name of the tool. - `description`: Optional, but strongly-recommended description of the tool. - `input_schema`: [JSON schema](https://json-schema.org/draft/2020-12) for the tool `input` shape that the model will produce in `tool_use` output content blocks. For example, if you defined `tools` as: ```json [ { "name": "get_stock_price", "description": "Get the current stock price for a given ticker symbol.", "input_schema": { "type": "object", "properties": { "ticker": { "type": "string", "description": "The stock ticker symbol, e.g. AAPL for Apple Inc." } }, "required": ["ticker"] } } ] ``` And then asked the model "What's the S&P 500 at today?", the model might produce `tool_use` content blocks in the response like this: ```json [ { "type": "tool_use", "id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "name": "get_stock_price", "input": { "ticker": "^GSPC" } } ] ``` You might then run your `get_stock_price` tool with `{"ticker": "^GSPC"}` as an input, and return the following back to the model in a subsequent `user` message: ```json [ { "type": "tool_result", "tool_use_id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "content": "259.75 USD" } ] ``` Tools can be used for workflows that include running client-side tools and functions, or more generally whenever you want the model to produce a particular JSON structure of output. See our [guide](https://docs.claude.com/en/docs/tool-use) for more details. top_k: Only sample from the top K options for each subsequent token. Used to remove "long tail" low probability responses. [Learn more technical details here](https://towardsdatascience.com/how-to-sample-from-language-models-682bceb97277). Recommended for advanced use cases only. You usually only need to use `temperature`. top_p: Use nucleus sampling. In nucleus sampling, we compute the cumulative distribution over all the options for each subsequent token in decreasing probability order and cut it off once it reaches a particular probability specified by `top_p`. You should either alter `temperature` or `top_p`, but not both. Recommended for advanced use cases only. You usually only need to use `temperature`. betas: Optional header to specify the beta version(s) you want to use. extra_headers: Send extra headers extra_query: Add additional query parameters to the request extra_body: Add additional JSON properties to the request timeout: Override the client-level default timeout for this request, in seconds """ ... @overload async def create( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, stream: bool, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> BetaMessage | AsyncStream[BetaRawMessageStreamEvent]: """ Send a structured list of input messages with text and/or image content, and the model will generate the next message in the conversation. The Messages API can be used for either single queries or stateless multi-turn conversations. Learn more about the Messages API in our [user guide](https://docs.claude.com/en/docs/initial-setup) Args: max_tokens: The maximum number of tokens to generate before stopping. Note that our models may stop _before_ reaching this maximum. This parameter only specifies the absolute maximum number of tokens to generate. Different models have different maximum values for this parameter. See [models](https://docs.claude.com/en/docs/models-overview) for details. messages: Input messages. Our models are trained to operate on alternating `user` and `assistant` conversational turns. When creating a new `Message`, you specify the prior conversational turns with the `messages` parameter, and the model then generates the next `Message` in the conversation. Consecutive `user` or `assistant` turns in your request will be combined into a single turn. Each input message must be an object with a `role` and `content`. You can specify a single `user`-role message, or you can include multiple `user` and `assistant` messages. If the final message uses the `assistant` role, the response content will continue immediately from the content in that message. This can be used to constrain part of the model's response. Example with a single `user` message: ```json [{ "role": "user", "content": "Hello, Claude" }] ``` Example with multiple conversational turns: ```json [ { "role": "user", "content": "Hello there." }, { "role": "assistant", "content": "Hi, I'm Claude. How can I help you?" }, { "role": "user", "content": "Can you explain LLMs in plain English?" } ] ``` Example with a partially-filled response from Claude: ```json [ { "role": "user", "content": "What's the Greek name for Sun? (A) Sol (B) Helios (C) Sun" }, { "role": "assistant", "content": "The best answer is (" } ] ``` Each input message `content` may be either a single `string` or an array of content blocks, where each block has a specific `type`. Using a `string` for `content` is shorthand for an array of one content block of type `"text"`. The following input messages are equivalent: ```json { "role": "user", "content": "Hello, Claude" } ``` ```json { "role": "user", "content": [{ "type": "text", "text": "Hello, Claude" }] } ``` See [input examples](https://docs.claude.com/en/api/messages-examples). Note that if you want to include a [system prompt](https://docs.claude.com/en/docs/system-prompts), you can use the top-level `system` parameter — there is no `"system"` role for input messages in the Messages API. There is a limit of 100,000 messages in a single request. model: The model that will complete your prompt.\n\nSee [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. stream: Whether to incrementally stream the response using server-sent events. See [streaming](https://docs.claude.com/en/api/messages-streaming) for details. container: Container identifier for reuse across requests. context_management: Context management configuration. This allows you to control how Claude manages context across multiple requests, such as whether to clear function results or not. mcp_servers: MCP servers to be utilized in this request metadata: An object describing metadata about the request. output_config: Configuration options for the model's output. Controls aspects like how much effort the model puts into its response. output_format: A schema to specify Claude's output format in responses. service_tier: Determines whether to use priority capacity (if available) or standard capacity for this request. Anthropic offers different levels of service for your API requests. See [service-tiers](https://docs.claude.com/en/api/service-tiers) for details. stop_sequences: Custom text sequences that will cause the model to stop generating. Our models will normally stop when they have naturally completed their turn, which will result in a response `stop_reason` of `"end_turn"`. If you want the model to stop generating when it encounters custom strings of text, you can use the `stop_sequences` parameter. If the model encounters one of the custom sequences, the response `stop_reason` value will be `"stop_sequence"` and the response `stop_sequence` value will contain the matched stop sequence. system: System prompt. A system prompt is a way of providing context and instructions to Claude, such as specifying a particular goal or role. See our [guide to system prompts](https://docs.claude.com/en/docs/system-prompts). temperature: Amount of randomness injected into the response. Defaults to `1.0`. Ranges from `0.0` to `1.0`. Use `temperature` closer to `0.0` for analytical / multiple choice, and closer to `1.0` for creative and generative tasks. Note that even with `temperature` of `0.0`, the results will not be fully deterministic. thinking: Configuration for enabling Claude's extended thinking. When enabled, responses include `thinking` content blocks showing Claude's thinking process before the final answer. Requires a minimum budget of 1,024 tokens and counts towards your `max_tokens` limit. See [extended thinking](https://docs.claude.com/en/docs/build-with-claude/extended-thinking) for details. tool_choice: How the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all. tools: Definitions of tools that the model may use. If you include `tools` in your API request, the model may return `tool_use` content blocks that represent the model's use of those tools. You can then run those tools using the tool input generated by the model and then optionally return results back to the model using `tool_result` content blocks. There are two types of tools: **client tools** and **server tools**. The behavior described below applies to client tools. For [server tools](https://docs.claude.com/en/docs/agents-and-tools/tool-use/overview#server-tools), see their individual documentation as each has its own behavior (e.g., the [web search tool](https://docs.claude.com/en/docs/agents-and-tools/tool-use/web-search-tool)). Each tool definition includes: - `name`: Name of the tool. - `description`: Optional, but strongly-recommended description of the tool. - `input_schema`: [JSON schema](https://json-schema.org/draft/2020-12) for the tool `input` shape that the model will produce in `tool_use` output content blocks. For example, if you defined `tools` as: ```json [ { "name": "get_stock_price", "description": "Get the current stock price for a given ticker symbol.", "input_schema": { "type": "object", "properties": { "ticker": { "type": "string", "description": "The stock ticker symbol, e.g. AAPL for Apple Inc." } }, "required": ["ticker"] } } ] ``` And then asked the model "What's the S&P 500 at today?", the model might produce `tool_use` content blocks in the response like this: ```json [ { "type": "tool_use", "id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "name": "get_stock_price", "input": { "ticker": "^GSPC" } } ] ``` You might then run your `get_stock_price` tool with `{"ticker": "^GSPC"}` as an input, and return the following back to the model in a subsequent `user` message: ```json [ { "type": "tool_result", "tool_use_id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "content": "259.75 USD" } ] ``` Tools can be used for workflows that include running client-side tools and functions, or more generally whenever you want the model to produce a particular JSON structure of output. See our [guide](https://docs.claude.com/en/docs/tool-use) for more details. top_k: Only sample from the top K options for each subsequent token. Used to remove "long tail" low probability responses. [Learn more technical details here](https://towardsdatascience.com/how-to-sample-from-language-models-682bceb97277). Recommended for advanced use cases only. You usually only need to use `temperature`. top_p: Use nucleus sampling. In nucleus sampling, we compute the cumulative distribution over all the options for each subsequent token in decreasing probability order and cut it off once it reaches a particular probability specified by `top_p`. You should either alter `temperature` or `top_p`, but not both. Recommended for advanced use cases only. You usually only need to use `temperature`. betas: Optional header to specify the beta version(s) you want to use. extra_headers: Send extra headers extra_query: Add additional query parameters to the request extra_body: Add additional JSON properties to the request timeout: Override the client-level default timeout for this request, in seconds """ ... @required_args(["max_tokens", "messages", "model"], ["max_tokens", "messages", "model", "stream"]) async def create( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: Literal[False] | Literal[True] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> BetaMessage | AsyncStream[BetaRawMessageStreamEvent]: validate_output_format(output_format) if not stream and not is_given(timeout) and self._client.timeout == DEFAULT_TIMEOUT: timeout = self._client._calculate_nonstreaming_timeout( max_tokens, MODEL_NONSTREAMING_TOKENS.get(model, None) ) if model in DEPRECATED_MODELS: warnings.warn( f"The model '{model}' is deprecated and will reach end-of-life on {DEPRECATED_MODELS[model]}.\nPlease migrate to a newer model. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.", DeprecationWarning, stacklevel=3, ) extra_headers = { **strip_not_given({"anthropic-beta": ",".join(str(e) for e in betas) if is_given(betas) else not_given}), **(extra_headers or {}), } return await self._post( "/v1/messages?beta=true", body=await async_maybe_transform( { "max_tokens": max_tokens, "messages": messages, "model": model, "container": container, "context_management": context_management, "mcp_servers": mcp_servers, "metadata": metadata, "output_config": output_config, "output_format": output_format, "service_tier": service_tier, "stop_sequences": stop_sequences, "stream": stream, "system": system, "temperature": temperature, "thinking": thinking, "tool_choice": tool_choice, "tools": tools, "top_k": top_k, "top_p": top_p, }, message_create_params.MessageCreateParamsStreaming if stream else message_create_params.MessageCreateParamsNonStreaming, ), options=make_request_options( extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout ), cast_to=BetaMessage, stream=stream or False, stream_cls=AsyncStream[BetaRawMessageStreamEvent], ) async def parse( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: Literal[False] | Literal[True] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> ParsedBetaMessage[ResponseFormatT]: if not stream and not is_given(timeout) and self._client.timeout == DEFAULT_TIMEOUT: timeout = self._client._calculate_nonstreaming_timeout( max_tokens, MODEL_NONSTREAMING_TOKENS.get(model, None) ) if model in DEPRECATED_MODELS: warnings.warn( f"The model '{model}' is deprecated and will reach end-of-life on {DEPRECATED_MODELS[model]}.\nPlease migrate to a newer model. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.", DeprecationWarning, stacklevel=3, ) betas = [beta for beta in betas] if is_given(betas) else [] if "structured-outputs-2025-11-13" not in betas: # Ensure structured outputs beta is included for parse method betas.append("structured-outputs-2025-11-13") extra_headers = { "X-Stainless-Helper": "beta.messages.parse", **strip_not_given({"anthropic-beta": ",".join(str(e) for e in betas) if is_given(betas) else NOT_GIVEN}), **(extra_headers or {}), } transformed_output_format: Optional[message_create_params.OutputFormat] | NotGiven = NOT_GIVEN if is_given(output_format) and output_format is not None: adapted_type: TypeAdapter[ResponseFormatT] = TypeAdapter(output_format) try: schema = adapted_type.json_schema() transformed_output_format = message_create_params.OutputFormat( schema=transform_schema(schema), type="json_schema" ) except pydantic.errors.PydanticSchemaGenerationError as e: raise TypeError( ( "Could not generate JSON schema for the given `output_format` type. " "Use a type that works with `pydanitc.TypeAdapter`" ) ) from e def parser(response: BetaMessage) -> ParsedBetaMessage[ResponseFormatT]: return parse_response( response=response, output_format=cast( ResponseFormatT, output_format if is_given(output_format) and output_format is not None else NOT_GIVEN, ), ) return await self._post( "/v1/messages?beta=true", body=maybe_transform( { "max_tokens": max_tokens, "messages": messages, "model": model, "container": container, "context_management": context_management, "mcp_servers": mcp_servers, "output_config": output_config, "metadata": metadata, "output_format": transformed_output_format, "service_tier": service_tier, "stop_sequences": stop_sequences, "stream": stream, "system": system, "temperature": temperature, "thinking": thinking, "tool_choice": tool_choice, "tools": tools, "top_k": top_k, "top_p": top_p, }, message_create_params.MessageCreateParamsNonStreaming, ), options=make_request_options( extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout, post_parser=parser, ), cast_to=cast(Type[ParsedBetaMessage[ResponseFormatT]], BetaMessage), stream=False, ) @overload def tool_runner( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, tools: Iterable[BetaAsyncRunnableTool], compaction_control: CompactionControl | Omit = omit, max_iterations: int | Omit = omit, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: Literal[False] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaAsyncToolRunner[ResponseFormatT]: ... @overload def tool_runner( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, tools: Iterable[BetaAsyncRunnableTool], compaction_control: CompactionControl | Omit = omit, stream: Literal[True], max_iterations: int | Omit = omit, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaAsyncStreamingToolRunner[ResponseFormatT]: ... @overload def tool_runner( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, tools: Iterable[BetaAsyncRunnableTool], compaction_control: CompactionControl | Omit = omit, stream: bool, max_iterations: int | Omit = omit, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaAsyncStreamingToolRunner[ResponseFormatT] | BetaAsyncToolRunner[ResponseFormatT]: ... def tool_runner( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, tools: Iterable[BetaAsyncRunnableTool], compaction_control: CompactionControl | Omit = omit, max_iterations: int | Omit = omit, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, stream: Literal[True] | Literal[False] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaAsyncToolRunner[ResponseFormatT] | BetaAsyncStreamingToolRunner[ResponseFormatT]: """Create a Message stream""" if model in DEPRECATED_MODELS: warnings.warn( f"The model '{model}' is deprecated and will reach end-of-life on {DEPRECATED_MODELS[model]}.\nPlease migrate to a newer model. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.", DeprecationWarning, stacklevel=3, ) extra_headers = { "X-Stainless-Helper": "BetaToolRunner", **strip_not_given({"anthropic-beta": ",".join(str(e) for e in betas) if is_given(betas) else NOT_GIVEN}), **(extra_headers or {}), } params = cast( message_create_params.ParseMessageCreateParamsBase[ResponseFormatT], { "max_tokens": max_tokens, "messages": messages, "model": model, "container": container, "context_management": context_management, "mcp_servers": mcp_servers, "metadata": metadata, "output_config": output_config, "output_format": output_format, "service_tier": service_tier, "stop_sequences": stop_sequences, "system": system, "temperature": temperature, "thinking": thinking, "tool_choice": tool_choice, "tools": [tool.to_dict() for tool in tools], "top_k": top_k, "top_p": top_p, }, ) if stream: return BetaAsyncStreamingToolRunner[ResponseFormatT]( tools=tools, params=params, options={ "extra_headers": extra_headers, "extra_query": extra_query, "extra_body": extra_body, "timeout": timeout, }, client=cast("AsyncAnthropic", self._client), max_iterations=max_iterations if is_given(max_iterations) else None, compaction_control=compaction_control if is_given(compaction_control) else None, ) return BetaAsyncToolRunner[ResponseFormatT]( tools=tools, params=params, options={ "extra_headers": extra_headers, "extra_query": extra_query, "extra_body": extra_body, "timeout": timeout, }, client=cast("AsyncAnthropic", self._client), max_iterations=max_iterations if is_given(max_iterations) else None, compaction_control=compaction_control if is_given(compaction_control) else None, ) def stream( self, *, max_tokens: int, messages: Iterable[BetaMessageParam], model: ModelParam, metadata: BetaMetadataParam | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[type[ResponseFormatT]] | Omit = omit, container: Optional[message_create_params.Container] | Omit = omit, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, service_tier: Literal["auto", "standard_only"] | Omit = omit, stop_sequences: SequenceNotStr[str] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, temperature: float | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[BetaToolUnionParam] | Omit = omit, top_k: int | Omit = omit, top_p: float | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN, ) -> BetaAsyncMessageStreamManager[ResponseFormatT]: if model in DEPRECATED_MODELS: warnings.warn( f"The model '{model}' is deprecated and will reach end-of-life on {DEPRECATED_MODELS[model]}.\nPlease migrate to a newer model. Visit https://docs.anthropic.com/en/docs/resources/model-deprecations for more information.", DeprecationWarning, stacklevel=3, ) extra_headers = { "X-Stainless-Helper-Method": "stream", "X-Stainless-Stream-Helper": "beta.messages", **strip_not_given({"anthropic-beta": ",".join(str(e) for e in betas) if is_given(betas) else NOT_GIVEN}), **(extra_headers or {}), } transformed_output_format: Optional[message_create_params.OutputFormat] | NotGiven = NOT_GIVEN if is_given(output_format) and output_format is not None: adapted_type: TypeAdapter[ResponseFormatT] = TypeAdapter(output_format) try: schema = adapted_type.json_schema() transformed_output_format = message_create_params.OutputFormat( schema=transform_schema(schema), type="json_schema" ) except pydantic.errors.PydanticSchemaGenerationError as e: raise TypeError( ( "Could not generate JSON schema for the given `output_format` type. " "Use a type that works with `pydanitc.TypeAdapter`" ) ) from e request = self._post( "/v1/messages", body=maybe_transform( { "max_tokens": max_tokens, "messages": messages, "model": model, "metadata": metadata, "output_config": output_config, "output_format": transformed_output_format, "container": container, "context_management": context_management, "mcp_servers": mcp_servers, "service_tier": service_tier, "stop_sequences": stop_sequences, "system": system, "temperature": temperature, "thinking": thinking, "top_k": top_k, "top_p": top_p, "tools": tools, "tool_choice": tool_choice, "stream": True, }, message_create_params.MessageCreateParams, ), options=make_request_options( extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout ), cast_to=BetaMessage, stream=True, stream_cls=AsyncStream[BetaRawMessageStreamEvent], ) return BetaAsyncMessageStreamManager(request, output_format=cast(ResponseFormatT, output_format)) async def count_tokens( self, *, messages: Iterable[BetaMessageParam], model: ModelParam, context_management: Optional[BetaContextManagementConfigParam] | Omit = omit, mcp_servers: Iterable[BetaRequestMCPServerURLDefinitionParam] | Omit = omit, output_config: BetaOutputConfigParam | Omit = omit, output_format: Optional[BetaJSONOutputFormatParam] | Omit = omit, system: Union[str, Iterable[BetaTextBlockParam]] | Omit = omit, thinking: BetaThinkingConfigParam | Omit = omit, tool_choice: BetaToolChoiceParam | Omit = omit, tools: Iterable[message_count_tokens_params.Tool] | Omit = omit, betas: List[AnthropicBetaParam] | Omit = omit, # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs. # The extra values given here take precedence over values defined on the client or passed to this method. extra_headers: Headers | None = None, extra_query: Query | None = None, extra_body: Body | None = None, timeout: float | httpx.Timeout | None | NotGiven = not_given, ) -> BetaMessageTokensCount: """ Count the number of tokens in a Message. The Token Count API can be used to count the number of tokens in a Message, including tools, images, and documents, without creating it. Learn more about token counting in our [user guide](https://docs.claude.com/en/docs/build-with-claude/token-counting) Args: messages: Input messages. Our models are trained to operate on alternating `user` and `assistant` conversational turns. When creating a new `Message`, you specify the prior conversational turns with the `messages` parameter, and the model then generates the next `Message` in the conversation. Consecutive `user` or `assistant` turns in your request will be combined into a single turn. Each input message must be an object with a `role` and `content`. You can specify a single `user`-role message, or you can include multiple `user` and `assistant` messages. If the final message uses the `assistant` role, the response content will continue immediately from the content in that message. This can be used to constrain part of the model's response. Example with a single `user` message: ```json [{ "role": "user", "content": "Hello, Claude" }] ``` Example with multiple conversational turns: ```json [ { "role": "user", "content": "Hello there." }, { "role": "assistant", "content": "Hi, I'm Claude. How can I help you?" }, { "role": "user", "content": "Can you explain LLMs in plain English?" } ] ``` Example with a partially-filled response from Claude: ```json [ { "role": "user", "content": "What's the Greek name for Sun? (A) Sol (B) Helios (C) Sun" }, { "role": "assistant", "content": "The best answer is (" } ] ``` Each input message `content` may be either a single `string` or an array of content blocks, where each block has a specific `type`. Using a `string` for `content` is shorthand for an array of one content block of type `"text"`. The following input messages are equivalent: ```json { "role": "user", "content": "Hello, Claude" } ``` ```json { "role": "user", "content": [{ "type": "text", "text": "Hello, Claude" }] } ``` See [input examples](https://docs.claude.com/en/api/messages-examples). Note that if you want to include a [system prompt](https://docs.claude.com/en/docs/system-prompts), you can use the top-level `system` parameter — there is no `"system"` role for input messages in the Messages API. There is a limit of 100,000 messages in a single request. model: The model that will complete your prompt.\n\nSee [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options. context_management: Context management configuration. This allows you to control how Claude manages context across multiple requests, such as whether to clear function results or not. mcp_servers: MCP servers to be utilized in this request output_config: Configuration options for the model's output. Controls aspects like how much effort the model puts into its response. output_format: A schema to specify Claude's output format in responses. system: System prompt. A system prompt is a way of providing context and instructions to Claude, such as specifying a particular goal or role. See our [guide to system prompts](https://docs.claude.com/en/docs/system-prompts). thinking: Configuration for enabling Claude's extended thinking. When enabled, responses include `thinking` content blocks showing Claude's thinking process before the final answer. Requires a minimum budget of 1,024 tokens and counts towards your `max_tokens` limit. See [extended thinking](https://docs.claude.com/en/docs/build-with-claude/extended-thinking) for details. tool_choice: How the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all. tools: Definitions of tools that the model may use. If you include `tools` in your API request, the model may return `tool_use` content blocks that represent the model's use of those tools. You can then run those tools using the tool input generated by the model and then optionally return results back to the model using `tool_result` content blocks. There are two types of tools: **client tools** and **server tools**. The behavior described below applies to client tools. For [server tools](https://docs.claude.com/en/docs/agents-and-tools/tool-use/overview#server-tools), see their individual documentation as each has its own behavior (e.g., the [web search tool](https://docs.claude.com/en/docs/agents-and-tools/tool-use/web-search-tool)). Each tool definition includes: - `name`: Name of the tool. - `description`: Optional, but strongly-recommended description of the tool. - `input_schema`: [JSON schema](https://json-schema.org/draft/2020-12) for the tool `input` shape that the model will produce in `tool_use` output content blocks. For example, if you defined `tools` as: ```json [ { "name": "get_stock_price", "description": "Get the current stock price for a given ticker symbol.", "input_schema": { "type": "object", "properties": { "ticker": { "type": "string", "description": "The stock ticker symbol, e.g. AAPL for Apple Inc." } }, "required": ["ticker"] } } ] ``` And then asked the model "What's the S&P 500 at today?", the model might produce `tool_use` content blocks in the response like this: ```json [ { "type": "tool_use", "id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "name": "get_stock_price", "input": { "ticker": "^GSPC" } } ] ``` You might then run your `get_stock_price` tool with `{"ticker": "^GSPC"}` as an input, and return the following back to the model in a subsequent `user` message: ```json [ { "type": "tool_result", "tool_use_id": "toolu_01D7FLrfh4GYq7yT1ULFeyMV", "content": "259.75 USD" } ] ``` Tools can be used for workflows that include running client-side tools and functions, or more generally whenever you want the model to produce a particular JSON structure of output. See our [guide](https://docs.claude.com/en/docs/tool-use) for more details. betas: Optional header to specify the beta version(s) you want to use. extra_headers: Send extra headers extra_query: Add additional query parameters to the request extra_body: Add additional JSON properties to the request timeout: Override the client-level default timeout for this request, in seconds """ extra_headers = { **strip_not_given( { "anthropic-beta": ",".join(chain((str(e) for e in betas), ["token-counting-2024-11-01"])) if is_given(betas) else not_given } ), **(extra_headers or {}), } extra_headers = {"anthropic-beta": "token-counting-2024-11-01", **(extra_headers or {})} return await self._post( "/v1/messages/count_tokens?beta=true", body=await async_maybe_transform( { "messages": messages, "model": model, "context_management": context_management, "mcp_servers": mcp_servers, "output_config": output_config, "output_format": output_format, "system": system, "thinking": thinking, "tool_choice": tool_choice, "tools": tools, }, message_count_tokens_params.MessageCountTokensParams, ), options=make_request_options( extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout ), cast_to=BetaMessageTokensCount, ) class MessagesWithRawResponse: def __init__(self, messages: Messages) -> None: self._messages = messages self.create = _legacy_response.to_raw_response_wrapper( messages.create, ) self.count_tokens = _legacy_response.to_raw_response_wrapper( messages.count_tokens, ) @cached_property def batches(self) -> BatchesWithRawResponse: return BatchesWithRawResponse(self._messages.batches) class AsyncMessagesWithRawResponse: def __init__(self, messages: AsyncMessages) -> None: self._messages = messages self.create = _legacy_response.async_to_raw_response_wrapper( messages.create, ) self.count_tokens = _legacy_response.async_to_raw_response_wrapper( messages.count_tokens, ) @cached_property def batches(self) -> AsyncBatchesWithRawResponse: return AsyncBatchesWithRawResponse(self._messages.batches) class MessagesWithStreamingResponse: def __init__(self, messages: Messages) -> None: self._messages = messages self.create = to_streamed_response_wrapper( messages.create, ) self.count_tokens = to_streamed_response_wrapper( messages.count_tokens, ) @cached_property def batches(self) -> BatchesWithStreamingResponse: return BatchesWithStreamingResponse(self._messages.batches) class AsyncMessagesWithStreamingResponse: def __init__(self, messages: AsyncMessages) -> None: self._messages = messages self.create = async_to_streamed_response_wrapper( messages.create, ) self.count_tokens = async_to_streamed_response_wrapper( messages.count_tokens, ) @cached_property def batches(self) -> AsyncBatchesWithStreamingResponse: return AsyncBatchesWithStreamingResponse(self._messages.batches) def validate_output_format(output_format: object) -> None: if inspect.isclass(output_format) and issubclass(output_format, pydantic.BaseModel): raise TypeError( "You tried to pass a `BaseModel` class to `beta.messages.create()`; You must use `beta.messages.parse()` instead" )