The model is currently live on the OpenRouter platform with a launch discount of 50%. Introductory pricing is $0.10 per million input tokens and $0.35 per million output tokens. Stefano Ermon described the release as the most intelligent diffusion LLM ever made available to the public.
Tuesday, October 20, 2026
Story
Inception's Mercury 3 Hits 1,000+ Tok/Sec to Beat GPT-6 Luna
AI1 day ago
A new artificial intelligence model from Inception offers a 1M token context window, representing a fourfold increase over the company's previous generation. Mercury 3, a diffusion LLM, processes data at speeds exceeding 1,000 tokens per second and is reported by the company to outperform GPT-6 Luna and Haiku 5.5 on quality. It specifically beats those models on the GPQA Diamond, IFBench, and SWE-Bench Atlas benchmarks.
Looking for more context or analysis?
Tilt chat provides reliable, tailored insights, alerts, and automations.