v0.16.0¶

Add OpenRouter as a thick provider, enabling access to 300+ AI models through a single unified API.

Highlights¶

OpenRouter Provider: Access Claude, GPT-4, Gemini, Llama, Mistral, and 300+ other models through OpenRouter's unified API
Full Feature Support: Streaming, tool calling, vision, and JSON mode
OAuth Authentication: Built-in PKCE flow for secure interactive authentication

Added¶

OpenRouter Thick Provider¶

The new omni-openrouter module provides SDK-based integration with OpenRouter:

import "github.com/plexusone/omnillm"

client, _ := omnillm.NewClient(omnillm.ClientConfig{
    Provider: omnillm.ProviderNameOpenRouter,
    APIKey:   os.Getenv("OPENROUTER_API_KEY"),
})

resp, _ := client.CreateChatCompletion(ctx, &omnillm.ChatCompletionRequest{
    Model: "anthropic/claude-3.5-sonnet",
    Messages: []omnillm.Message{
        {Role: omnillm.RoleUser, Content: "Hello!"},
    },
})

Provider Constant¶

New ProviderNameOpenRouter constant for configuration:

omnillm.ProviderNameOpenRouter // "openrouter"

Feature Support¶

Feature	Supported
Streaming	Yes
Tool Calling	Yes
Vision	Yes
JSON Mode	Yes
System Role	Yes

Available Models¶

OpenRouter provides access to models from multiple providers:

Anthropic: claude-opus-4, claude-sonnet-4, claude-3.5-sonnet, etc.
OpenAI: gpt-4-turbo, gpt-4o, gpt-3.5-turbo, etc.
Google: gemini-pro, gemini-1.5-pro, etc.
Meta: llama-3-70b, llama-3.1-405b, etc.
Mistral: mixtral-8x22b, mistral-large, etc.
And 300+ more

Model names use the format provider/model-name (e.g., anthropic/claude-3.5-sonnet).

Dependencies¶

github.com/plexusone/omni-openrouter v0.1.0

Installation¶

go get github.com/plexusone/omnillm@v0.16.0

Requirements¶

Go 1.26+