Introducing the OpenRouter MCP, live model intelligence right inside your agent Your agent builds and ships, but when it comes to choosing the right model for the right job, it guesses from 6 month old training data Watch it pick, price, and test the right model:

Connect it and your agent gets live intel: a 400+ model catalog, benchmarks, per-provider pricing & latency, test inference, and docs search Ask "top coding models under $2/M?", the most popular image model, or run a real prompt to compare answers+cost, w/o ever leaving your CLI

Pick, price, test, and wire in a model in under a minute Connect in two commands: claude mcp add --transport http openrouter claude mcp login openrouter Works with all your favorite agents Available now on

The model performance rankings come from our new Benchmarks API, allowing your agent to query live benchmark scores (incl Artificial Analysis and Design Arena) Fun result: @Zai_org’s GLM-5.2 is the best available model for both coding & design Docs:

OR should have a CLI as well, just saying

This is actually a pretty fkn big deal: 1. Add OpenRouter MCP to Codex as a streamable HTTP MCP server: 2. Authenticate through OpenRouter OAuth. OpenRouter says the MCP key defaults to a 7-day expiry and a $10 spend cap. 3. Wrap it in a small Codex plugin/skill so you can invoke it with @openrouter or similar. 4. Add skill behavior like: “use model anthropic/claude-opus-4.8” “use the cheapest strong model” “use the best model for code review” “compare GPT-5.5, Opus, Gemini, and DeepSeek” “recommend the best model, then run the prompt through it” 5. Enjoy 2x better Codex.

@OpenRouter can you please provide the range of prices and not only the cheapest one for a given model? 🙏

🔥 OpenRouter MCP is changing the game. Now your agent can pick, compare, price, and test models in real time instead of relying on outdated assumptions.

yo @jjacky where'd you get this sweater it looks v comfy. Also huge congrats on the release!!!

Really good job I needed this one

Oooooh very cool

@NousResearch

honestly, great tools and research that openrouter is putting out. good job guys 🫡

😍😍😍

you can also give your agent access to web search to give it real-time data

Great news! The number of times I let my agents read your docs was insane. 😄

Upvote this.

Interesting! Does this work with BYOK?

if openrouter already provides live model data why would an agent still be guessing from six month old training data.

That's a good one 👏

wow @tweem 500k views

big launch energy. gg

release a batch api you lazy bastards, i don't need these side quest features

This is a very interesting development! Congrats on shipping this, team

This model is now brining a new development just like MultichainZ AI studio

@tweem 100k views

thats sick 🤝

这下 agent 不用凭记忆选模型了

super helpful

you should add more niche benchmarks in like deepswe the well known ones are being benchmaxed

Mungu utanimaliza na baraka

Banger

cheeky

@KenTheRogers 👀

Love it. What about skills so that AI can write better AI prompts? Been struggling with this

Huge

Thank you OpenRouter! Game Changer.

Actual game changer I've been wrestling with picking the right agents for the job