API Compatibility

Anthropic · Messages

Generates a message completion using the Anthropic Messages API shape.#

POST/v1/messages

Anthropic-compatible Messages endpoint. Designed to be a drop-in target for clients that use the Anthropic SDK (including Claude Code via ANTHROPIC_BASE_URL). Supports text, vision (image content blocks), tool use, streaming (SSE) and non-streaming responses.

Reference: https://docs.anthropic.com/en/api/messages

Request body

application/json ·

Property	Type	Description
`model`	string	The model that will complete the prompt. When omitted or empty, the server falls back to its configured default chat model.
`messages`	[]	Input messages. Roles alternate "user" and "assistant". Each may have string or content-block content.
`max_tokens`	object (int32)	Maximum number of tokens to generate before stopping.
`system`	object	System prompt, supplied as a string or an array of system text blocks.
`metadata`	object	Free-form request metadata. user_id is the canonical field. Stored, not interpreted.
`stop_sequences`	string[]	Custom stop sequences. Generation stops when any of these appears in the output.
`stream`	boolean	When true, the response is sent as a server-sent event stream.
`temperature`	object (double)	Sampling temperature in [0, 1]. Higher = more random.
`top_p`	object (double)	Nucleus sampling cutoff. Only sample tokens within the top_p probability mass.
`top_k`	object (int32)	Only sample from the top K tokens. Currently not honored by the LM-Kit backend (use temperature/top_p).
`tools`	[]	Tools the model may call.
`tool_choice`	object	How the model should choose tools. Accepts { "type": "auto" \| "any" \| "none" \| "tool", "name": "..." }.
`service_tier`	string	Service tier hint. Accepted and stored; not interpreted.
`thinking`		Extended-thinking configuration: { "type": "enabled", "budget_tokens": N } or { "type": "disabled" }. Disabled turns reasoning off for the request; enabled keeps the model's own reasoning level (the budget rides within max_tokens).

Responses

Status	Type	Description
200		OK
400		Bad Request
404		Not Found
500	application/json	Internal Server Error

curl -X POST "$LMKIT_ONE_URL/v1/messages" \
  -H "Authorization: Bearer $LMKIT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "string",
  "messages": [
    {
      "role": "string",
      "content": {}
    }
  ],
  "max_tokens": "string",
  "system": {}
}'