Capability · Framework — orchestration

OpenRouter

OpenRouter is the easiest way to evaluate and mix many LLMs without signing up with every vendor. You top up credits once, and their unified endpoint routes requests to whichever model you pick — with live pricing, automatic fallbacks across providers, and a leaderboard of real-world usage that's become a de-facto benchmark of which models developers actually use.

Framework facts

Category: orchestration
Language: API (any client)
License: commercial
Repository: https://github.com/OpenRouterTeam

Install

# Works with any OpenAI-compatible SDK
pip install openai
# or
npm install openai

Quickstart

from openai import OpenAI

client = OpenAI(
    base_url='https://openrouter.ai/api/v1',
    api_key='sk-or-...'
)
resp = client.chat.completions.create(
    model='anthropic/claude-opus-4-7',
    messages=[{'role': 'user', 'content': 'Explain MCP'}]
)
print(resp.choices[0].message.content)

Alternatives

LiteLLM — self-hosted SDK/proxy equivalent
Portkey — adds guardrails and caching
Together AI — hosts open-source models with its own router
Requesty — newer AI request router

Frequently asked questions

Is OpenRouter cheaper than going direct?

For most models OpenRouter prices match the upstream provider, and they pass through promotions. You pay a small markup on some routes in exchange for unified billing, automatic fallback, and zero per-vendor signup friction.

Can I use it from agent frameworks?

Yes. Because it speaks OpenAI's API, LangChain, LlamaIndex, CrewAI, LangGraph, DSPy, and most other frameworks work with OpenRouter by setting the base URL and a model string like 'anthropic/claude-opus-4-7'.

Sources

OpenRouter — docs — accessed 2026-04-20
OpenRouter model list — accessed 2026-04-20