OpenRouter Provider

Access 200+ AI models through a single API

OpenRouter is an AI model aggregator that provides unified access to 200+ models from OpenAI, Anthropic, Google, Meta, Mistral, and many others.

Configuration

provider · string · required

Set to "openrouter"

api_key · string · required

OpenRouter API key. Get yours at openrouter.ai/keys

model · string · required

Fully qualified model name in format provider/organization/model-name

temperature · number

Sampling temperature (0.0-2.0). Defaults to 0.7.

max_tokens · number

Maximum output tokens.

Example Configuration

{
  "provider": "openrouter",
  "model": "openrouter/anthropic/claude-sonnet-4",
  "api_key": "sk-or-v1-...",
  "temperature": 0.7
}

Model Selection

OpenRouter uses fully qualified model names:

• openrouter/anthropic/claude-sonnet-4
• openrouter/openai/gpt-4o
• openrouter/google/gemini-2.0-flash
• openrouter/meta-llama/llama-3.2-90b-vision
• openrouter/mistralai/mistral-large-2411

Browse available models at openrouter.ai/models

Authentication

OpenRouter uses Bearer token authentication with additional headers:

Authorization: Bearer sk-or-v1-...
HTTP-Referer: https://github.com/nullclaw/nullclaw
X-Title: nullclaw

The HTTP-Referer and X-Title headers are automatically added by the provider.

Capabilities

Feature	Support
Streaming	Yes
Function Calling	Yes (model-dependent)
Vision (images)	Yes (model-dependent)
System Messages	Yes
Tool Calls	Yes

Reasoning Models

OpenRouter supports OpenAI reasoning models (o1, o3, gpt-5):

{
  "provider": "openrouter",
  "model": "openrouter/openai/o3-mini",
  "api_key": "sk-or-v1-...",
  "max_tokens": 100,
  "reasoning_effort": "medium"
}

• Temperature is automatically omitted for reasoning models
• Uses max_completion_tokens field
• Supports reasoning_effort parameter

Warmup

OpenRouter provider includes a warmup feature to pre-establish TLS connections:

pub fn warmup(self: *OpenRouterProvider) void {
    const api_key = self.api_key orelse return;
    // Hit auth endpoint to warm up connection
    const resp = curlGet(self.allocator, WARMUP_URL, auth_hdr) catch return;
    self.allocator.free(resp);
}

The warmup hits https://openrouter.ai/api/v1/auth/key to establish a connection.

Code Example

From src/providers/openrouter.zig:

pub const OpenRouterProvider = struct {
    api_key: ?[]const u8,
    allocator: std.mem.Allocator,

    const BASE_URL = "https://openrouter.ai/api/v1/chat/completions";
    const WARMUP_URL = "https://openrouter.ai/api/v1/auth/key";
    const REFERER = "https://github.com/nullclaw/nullclaw";
    const TITLE = "nullclaw";

    pub fn init(allocator: std.mem.Allocator, api_key: ?[]const u8) OpenRouterProvider {
        return .{
            .api_key = api_key,
            .allocator = allocator,
        };
    }
};

Multi-Turn Conversations

OpenRouter supports full conversation history:

{
  "model": "openrouter/anthropic/claude-sonnet-4",
  "messages": [
    {"role": "system", "content": "You are a helpful assistant"},
    {"role": "user", "content": "Hello"},
    {"role": "assistant", "content": "Hi! How can I help?"},
    {"role": "user", "content": "What's 2+2?"}
  ]
}

Error Handling

The provider classifies common OpenRouter API errors:

• error.RateLimited — 429 rate limit exceeded
• error.ContextLengthExceeded — Context window too large
• error.InvalidApiKey — Authentication failed
• error.ApiError — Generic API error
• error.NoResponseContent — Empty response

Pricing

OpenRouter charges based on the model you use. Pricing varies by model:

• Check current pricing at openrouter.ai/models
• Set spending limits in your OpenRouter dashboard
• Monitor usage at openrouter.ai/activity

Related

• Providers Overview
• OpenAI Provider
• Anthropic Provider