Our load balancer is providing a 5 to 20% uptime boost for OpenAI's new gpt-oss-120b

Let's compare OpenAI gpt-oss and Qwen-3 on maths & reasoning:

Today, OpenAI returns to open source! 🚀 Excited to be a launch partner for gpt-oss, OpenAI’s new open-weight models with variable reasoning. First providers are @FireworksAI_HQ, @GroqInc, and @CerebrasSystems, with more to come! Here’s what's launching today 👇

Another handy new feature for LLM app developers: PDF URLs You can now submit PDF URLs directly to OpenRouter without having to encode them to base64. And it works with ANY model.

NEW: Audio Inputs🔈 Let users talk to your app. See all LLMs that support audio and sample code below 👇

Notice to users of our Chatroom: we now have a default system message that 1) helps with markdown formatting and 2) helps models identify themselves by their name and version, to help with identity confusion (models don’t know who they are from their training data). The API is unaffected, and you can view and change the system prompt in the Chatroom whenever you want.

A New stealth model: Horizon Beta on @OpenRouterAI is now available in anycoder for vibe coding An improved version of Horizon Alpha Try it out on your phone

New stealth model: Horizon Beta 🌄 An improved version of Horizon Alpha.

Horizon Alpha, a new stealth model on @OpenRouterAI is now available in anycoder for vibe coding

Horizon-alpha, the new stealthed model on openrouter, just swept the EQ-Bench leaderboards. Links & writing samples >>