Inception Mercury 3 claims speed edge over rivals

Inception AI released Mercury 3, its new flagship diffusion language model, claiming it beat GPT-6 Luna and Claude Haiku 5.5 on eight of 12 benchmarks at four to eight times their speed. Mercury 3 reached 1,068 tokens per second versus roughly 240 for Haiku 5.5 and 130 for GPT-6 Luna. Its biggest lead came in coding, scoring 51.1 on SWE-Bench Atlas against 33.3 and 32.3 for rivals. The model supports a 1 million token context window and is priced at 20 cents per million input tokens and 70 cents per million output tokens, with a launch promotion halving both rates.