Chat APIs

FIM Gate provides both the chat/completions and responses OpenAI-format APIs. This page explains where each one applies.

responses API

All models from non-OpenAI providers can be called through the responses API. For OpenAI's own models, responses support follows the official OpenAI model list.

chat API

Some OpenAI models natively offer only the Responses API. So you don't have to rework an existing chat/completions integration, FIM Gate provides chat-API translation for the following models:

• o3-pro / o3-pro-2025-06-10
• o1-pro / o1-pro-2025-03-19
• o3-deep-research / o3-deep-research-2025-06-26
• o4-mini-deep-research / o4-mini-deep-research-2025-06-26
• codex-mini-latest

For these models, send requests to https://api.gate.fim.ai/v1/chat/completions as usual, and use the extended capabilities below.

Calling built-in tools

Declare a tool type in the tools array to enable the model's built-in tools, e.g. web search:

{
  "model": "o3-pro",
  "stream": true,
  "messages": [
    {"role": "user", "content": "Look up the release notes for the latest Rust version"}
  ],
  "tools": [
    {"type": "web_search_preview"}
  ]
}

Getting a reasoning summary

Request a reasoning summary via reasoning.summary; the result appears in the reasoning_content field of the response message:

{
  "model": "o3-pro",
  "stream": true,
  "messages": [
    {"role": "user", "content": "Prove that the square root of 2 is irrational"}
  ],
  "reasoning": {
    "summary": "detailed"
  }
}

Setting reasoning effort

The following two forms are equivalent — use either one:

{
  "model": "o3-pro",
  "stream": true,
  "messages": [
    {"role": "user", "content": "Prove that the square root of 2 is irrational"}
  ],
  "reasoning": {
    "effort": "high"
  }
}

{
  "model": "o3-pro",
  "stream": true,
  "messages": [
    {"role": "user", "content": "Prove that the square root of 2 is irrational"}
  ],
  "reasoning_effort": "high"
}