Diffusion LLM

Mercury 2.5 Says It Is 40 Percent Smarter and Publishes No Benchmark to Prove It

Inception's Mercury 2.5 claims 1,107 tokens per second and a 40 percent jump in intelligence. We pulled OpenRouter's live model catalog to check: the 260,000 token context and the pricing are exactly as advertised, and there is not a single benchmark score behind the quality claim.