FIM Gate: Every AI Model, One API

When your request/response fields must match the official Anthropic API exactly (for example, to reuse existing code built on the Anthropic SDK), use FIM Gate's Claude native entry point. If you don't need field-level compatibility, calling Claude models through the OpenAI-compatible API is simpler.

Endpoint

Item	Value
Base URL	`https://api.gate.fim.ai`
Method and path	`POST /v1/messages`
Authentication	`x-api-key: $FIM_API_KEY`

Note that the auth header is x-api-key, not Authorization: Bearer. We also recommend sending the anthropic-version: 2023-06-01 header, matching how the official Anthropic API is called.

Request parameters

Parameter	Type	Required	Description
`model`	string	Yes	Model ID, e.g. `claude-sonnet-4-5`
`max_tokens`	integer	Yes	Output token limit. This field is required in the Messages API
`messages`	array	Yes	Message array; `role` is limited to `user` / `assistant` and must alternate
`system`	string / array	No	System prompt, passed separately rather than inside `messages`
`stream`	boolean	No	Set to `true` to enable SSE streaming
`temperature`	number	No	Sampling temperature, `0`–`1`
`top_p`	number	No	Nucleus sampling threshold
`top_k`	integer	No	Sample only from the K highest-probability candidates
`stop_sequences`	array	No	Custom stop sequences
`tools`	array	No	Tool definitions, each with `name`, `description`, and `input_schema` (JSON Schema)
`metadata`	object	No	Request metadata, e.g. `user_id`

Two key differences from the OpenAI format: max_tokens is required, and the system prompt goes in a separate system field.

curl example

curl https://api.gate.fim.ai/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: $FIM_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -d '{
    "model": "claude-sonnet-4-5",
    "max_tokens": 1024,
    "system": "You are a rigorous technical assistant",
    "messages": [
      {"role": "user", "content": "Explain the TCP three-way handshake"}
    ]
  }'

Response structure:

{
  "id": "msg_abc123",
  "type": "message",
  "role": "assistant",
  "model": "claude-sonnet-4-5",
  "content": [
    {
      "type": "text",
      "text": "The TCP three-way handshake is the process of establishing a connection..."
    }
  ],
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 32,
    "output_tokens": 215
  }
}

The generated text is in the text blocks of the content array; tool calls appear as type: "tool_use" blocks.

stop_reason values

Value	Meaning
`end_turn`	The model finished naturally
`max_tokens`	Output was truncated at the `max_tokens` limit
`stop_sequence`	A sequence in `stop_sequences` was matched
`tool_use`	The model requested a tool call

Streaming

Add "stream": true to the request body and the response becomes an SSE event stream. Events arrive in this order: message_start, content_block_start, a series of content_block_delta (incremental text in delta.text), content_block_stop, message_delta, and message_stop.

Using the Anthropic SDK

pip install anthropic

import anthropic

client = anthropic.Anthropic(
    api_key="$FIM_API_KEY",
    base_url="https://api.gate.fim.ai",
)

message = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Hello, Claude"}
    ],
)
print(message.content)

The SDK automatically appends /v1/messages to base_url, so base_url should stop at the domain.

Claude Native API