Messages API
A conversational interface provided by Anthropic that models input as a list of role-based messages, expressing model behavior uniformly through content blocks and stop_reason. It is stateless, so the full history must be sent with each request.
Timeline (last 90 days)
Anthropic's Messages API is stateless, so every request must resend the full conversation history.
If the project makes heavy use of Text Completions or temperature, the migration cost is relatively high and you can wait for a more complete transition plan; if it only makes standard Messages API calls, the upgrade path is relatively smooth.
The Text Completions interface accepts a single prompt and returns a continuation; the Messages API request body consists of a system prompt and a messages array, with each message containing a role and content.
Anthropic has fully consolidated conversational capabilities into the Messages API, and the old text completion endpoint is no longer supported.
The Anthropic Messages API stream carries the input token count in the message_start event (message.usage), and output tokens arrive later via the message_delta event.
The usage field in the Messages API has always been used to return the token statistics of the current call (such as input_tokens, output_tokens).
In Anthropic SDK v1.15.0, the Messages API adds a speed parameter, and a usage.speed field is added to the response body.
OpenAI offers hosted file search (with managed storage) and a complete real-time voice model line; Anthropic offers server-side web search, web fetching, code execution, a tool search tool for large tool sets, and a connector that can connect to remote MCP servers directly from the Messages API.
Anthropic's Messages API is a stateless conversational interface that models input as a list of role-based messages, expresses the reason for stopping in the response via the stop_reason field on content blocks, and requires the full history to be included with each request.
The Messages API is the interface for model inference, through which developers send conversation requests and obtain completion results; the Admin API is the interface for organization management, used to query operational data such as members, keys, and usage reports.
6 more timeline events
All Facts (17)
Anthropic使用Messages API并引入了独特的system prompt处理方式
90%UnverifiedAnthropic's Messages API is stateless, so every request must resend the full conversation history.
50%UnverifiedIf the project makes heavy use of Text Completions or temperature, the migration cost is relatively high and you can wait for a more complete transition plan; if it only makes standard Messages API calls, the upgrade path is relatively smooth.
50%UnverifiedAnthropic has fully consolidated conversational capabilities into the Messages API, and the old text completion endpoint is no longer supported.
50%UnverifiedThe Text Completions interface accepts a single prompt and returns a continuation; the Messages API request body consists of a system prompt and a messages array, with each message containing a role and content.
50%UnverifiedThe Anthropic Messages API stream carries the input token count in the message_start event (message.usage), and output tokens arrive later via the message_delta event.
50%UnverifiedIn Anthropic SDK v1.15.0, the Messages API adds a speed parameter, and a usage.speed field is added to the response body.
50%UnverifiedThe usage field in the Messages API has always been used to return the token statistics of the current call (such as input_tokens, output_tokens).
50%UnverifiedAnthropic's Messages API is a stateless conversational interface that models input as a list of role-based messages, expresses the reason for stopping in the response via the stop_reason field on content blocks, and requires the full history to be included with each request.
50%UnverifiedOpenAI offers hosted file search (with managed storage) and a complete real-time voice model line; Anthropic offers server-side web search, web fetching, code execution, a tool search tool for large tool sets, and a connector that can connect to remote MCP servers directly from the Messages API.
50%UnverifiedDuring the above outage, Anthropic's Messages API requests were unaffected, and model inference itself continued to run as usual.
50%UnverifiedThe Messages API and the Admin API belong to different service boundaries within Anthropic's architecture, with independent rate limiting, authentication, and backend dependencies.
50%UnverifiedThe usage query path additionally depends on the log pipeline, statistics storage and report query layer, whereas the Messages API only needs the inference cluster and gateway health to work.
50%UnverifiedThe Messages API is the interface for model inference, through which developers send conversation requests and obtain completion results; the Admin API is the interface for organization management, used to query operational data such as members, keys, and usage reports.
50%UnverifiedThe script called the Opus 4.6 model via the standard Messages API, because the developer had copied the configuration from another project and forgotten to change the model name.
50%UnverifiedAnthropic updated the Messages API to restrict editing of thinking blocks in prior Claude context during multi-turn conversations, with the official rationale being to increase the difficulty of distillation attacks.
50%UnverifiedThe Anthropic-compatible interface follows the Messages API specification, with the endpoint at /v1/messages; the system prompt is a top-level independent field, and in the response structure, content is an array of objects containing type and text.
50%