Inception AI Launches Mercury 3 Diffusion Model on OpenRouter

Inception AI's new diffusion model, Mercury 3, is now live on OpenRouter. According to the listing, the model generates more than 1,000 tokens per second, supports a 1 million-token context window, and delivers 4x the performance of Mercury 2.5. Inception says Mercury 3 outperforms GPT-6 Luna and Haiku 5.5 on the IFBench, GPQA Diamond, and SWE-Bench Atlas benchmarks. At launch, the model is offered at a 50% discount, with pricing set at $0.10 per million input tokens and $0.35 per million output tokens.