API Reference

Quick Start

Set two environment variables and you're done:

export ANTHROPIC_BASE_URL=https://openlimits.app
export ANTHROPIC_API_KEY=your-key-here

That's it. Any Anthropic-compatible client (Claude Code CLI, Conductor, OpenCode, Cursor, etc.) will route through OpenLimits automatically.

Authentication

All API requests require authentication via one of:

• x-api-key: YOUR_KEY header (Anthropic style)
• Authorization: Bearer YOUR_KEY header (OpenAI style)

curl https://openlimits.app/v1/messages \
  -H "x-api-key: YOUR_KEY" \
  -H "content-type: application/json" \
  -d '{"model":"anthropic/claude-opus-4.8","max_tokens":1024,"messages":[{"role":"user","content":"Hello"}]}'

POST /v1/messages

Proxies directly to the Anthropic Messages API. The request and response formats are identical — see Anthropic's docs for the full schema.

Request

POST /v1/messages
Content-Type: application/json
x-api-key: YOUR_KEY

{
  "model": "anthropic/claude-opus-4.8",
  "max_tokens": 1024,
  "messages": [
    { "role": "user", "content": "Explain recursion in one sentence." }
  ],
  "stream": false
}

Response

{
  "id": "msg_...",
  "type": "message",
  "role": "assistant",
  "content": [
    { "type": "text", "text": "Recursion is when a function calls itself..." }
  ],
  "model": "anthropic/claude-opus-4.8",
  "usage": {
    "input_tokens": 14,
    "output_tokens": 32,
    "cache_read_input_tokens": 0,
    "cache_creation_input_tokens": 0
  }
}

Streaming

Set "stream": true to receive server-sent events (SSE). The event format follows the Anthropic streaming spec.

Effort Levels

Control quality vs speed with the effort parameter:

{
  "model": "anthropic/claude-opus-4.8",
  "max_tokens": 1024,
  "messages": [...],
  "output_config": { "effort": "low" }
}

Valid values: low, medium, high.

Extended Thinking

Supported via the anthropic-beta: interleaved-thinking-2025-05-14 header. See Anthropic's extended thinking docs.

POST /v1/responses

Proxies to the OpenAI Responses API for GPT models. This is the endpoint used by Codex CLI and Codex Desktop when configured with OpenLimits.

Request

POST /v1/responses
Content-Type: application/json
Authorization: Bearer YOUR_KEY

{
  "model": "openai/gpt-5.5",
  "instructions": "You are a helpful assistant.",
  "input": "Explain recursion in one sentence.",
  "stream": true,
  "store": false
}

Response

Returns a server-sent event (SSE) stream. Streaming is always enabled for this endpoint. The event format follows the OpenAI Responses API spec.

Supported Models

Use the OpenRouter-style OpenAI model ID for GPT-5.5:

• openai/gpt-5.5

Notes

• stream is always set to true (required by the upstream provider)
• store is always set to false (required by the upstream provider)
• If instructions is not provided, a default is used

POST /v1/chat/completions

OpenAI-compatible endpoint. Send requests in the OpenAI format and we translate to Anthropic on the fly. Use this with any OpenAI SDK client.

Request

POST /v1/chat/completions
Content-Type: application/json
Authorization: Bearer YOUR_KEY

{
  "model": "anthropic/claude-opus-4.8",
  "messages": [
    { "role": "system", "content": "You are a helpful assistant." },
    { "role": "user", "content": "Hello" }
  ],
  "max_tokens": 1024,
  "stream": false
}

Response

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "created": 1700000000,
  "model": "anthropic/claude-opus-4.8",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "Hello! How can I help?" },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 20,
    "completion_tokens": 8,
    "total_tokens": 28
  }
}

Model Routing

Claude model names are translated to the Anthropic Messages API format automatically. For OpenAI GPT models, use /v1/responses instead.

GET /v1/models

Returns a list of all available models in OpenAI-compatible format.

GET /v1/models
x-api-key: YOUR_KEY

{
  "object": "list",
  "data": [
    { "id": "anthropic/claude-opus-4.8", "object": "model", "owned_by": "anthropic" },
    { "id": "anthropic/claude-opus-4.7", "object": "model", "owned_by": "anthropic" },
    { "id": "anthropic/claude-sonnet-4.6", "object": "model", "owned_by": "anthropic" },
    { "id": "openai/gpt-5.5", "object": "model", "owned_by": "openai" },
    { "id": "z-ai/glm-5.1", "object": "model", "owned_by": "zai" },
    { "id": "z-ai/glm-5-turbo", "object": "model", "owned_by": "zai" },
    { "id": "minimax/minimax-m3", "object": "model", "owned_by": "minimax" },
    { "id": "deepseek/deepseek-v4-pro", "object": "model", "owned_by": "deepseek" },
    { "id": "deepseek/deepseek-v4-flash", "object": "model", "owned_by": "deepseek" }
  ]
}

Supported Models

Claude (Anthropic)

Available via /v1/messages and /v1/chat/completions. Use the same model IDs as OpenRouter:

GPT (OpenAI)

Available via /v1/responses:

Open Model Groups

Available on plans that include open model access. Use OpenRouter-style IDs:

Errors

Errors follow the Anthropic error format:

{
  "type": "error",
  "error": {
    "type": "authentication_error",
    "message": "Invalid or disabled authentication token"
  }
}

Client Setup Examples

Claude Code CLI

Run the automated setup script to configure Claude Code and Codex CLI in one step, or add manually to ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://openlimits.app",
    "ANTHROPIC_AUTH_TOKEN": "your-key-here"
  }
}

Conductor

Conductor automatically uses your Claude Code CLI settings (if installed). Otherwise, to force Conductor to use our API, set these environment variables:

Settings → Env

ANTHROPIC_BASE_URL=https://openlimits.app
ANTHROPIC_API_KEY=your-key-here

Python (Anthropic SDK)

import anthropic

client = anthropic.Anthropic(
    api_key="your-key-here",
    base_url="https://openlimits.app",
)

message = client.messages.create(
    model="anthropic/claude-opus-4.8",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}],
)
print(message.content[0].text)

Python (OpenAI SDK)

from openai import OpenAI

client = OpenAI(
    api_key="your-key-here",
    base_url="https://openlimits.app/v1",
)

response = client.chat.completions.create(
    model="anthropic/claude-opus-4.8",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

TypeScript (Anthropic SDK)

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: "your-key-here",
  baseURL: "https://openlimits.app",
});

const message = await client.messages.create({
  model: "anthropic/claude-opus-4.8",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello" }],
});
console.log(message.content[0].text);

cURL

curl https://openlimits.app/v1/messages \
  -H "x-api-key: your-key-here" \
  -H "content-type: application/json" \
  -d '{
    "model": "anthropic/claude-opus-4.8",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Codex CLI / Desktop

OpenLimits works as a drop-in backend for Codex CLI and Codex Desktop. The setup script configures everything automatically, or you can set it up manually below.

Codex CLI (Recommended: config.toml)

Create or edit ~/.codex/config.toml with a custom provider:

# ~/.codex/config.toml
model = "openai/gpt-5.5"
model_provider = "openlimits"

[model_providers.openlimits]
name = "OpenLimits"
base_url = "https://openlimits.app/v1"
env_key = "OPENAI_API_KEY"
wire_api = "responses"

Then set the API key in your shell profile (~/.zshrc or ~/.bashrc):

export OPENAI_API_KEY=your-key-here

Important: Remove OPENAI_BASE_URL and CODEX_OPENAI_BASE_URL from your environment if set — they override the config.toml settings.

Codex Desktop

Open Codex Desktop settings and set:

Model Selection

With OpenLimits, use the OpenRouter-style GPT-5.5 model ID in Codex CLI:

codex --model openai/gpt-5.5

Environment Variables

A reference of all environment variables you can use to configure clients with OpenLimits.

Anthropic-compatible clients

For Claude Code CLI, Conductor, and any Anthropic SDK client:

OpenAI-compatible clients

For Codex CLI/Desktop, Cursor, and any OpenAI SDK client:

Note: Anthropic clients use https://openlimits.app (no /v1), while OpenAI clients use https://openlimits.app/v1 (with /v1). This is because the Anthropic SDK appends /v1/messages automatically, while the OpenAI SDK appends only the endpoint path.

Quick copy

# For Claude Code / Anthropic clients
export ANTHROPIC_BASE_URL=https://openlimits.app
export ANTHROPIC_API_KEY=your-key-here

# For Codex / OpenAI clients
export OPENAI_BASE_URL=https://openlimits.app/v1
export OPENAI_API_KEY=your-key-here