Inception's Mercury 3 beats GPT-6 Luna, Claude Haiku 5.5 in speed
inception releases mercury 3, a diffusion-based language model that generates over 1,000 tokens per second, beating gpt-6 luna and claude haiku 5.5 in eight of 12 tests. this matters for real-time apps like ai agents and customer support.
inception has released mercury 3, a diffusion-based language model that generates over 1,068 tokens per second on nvidia gpus. in internal benchmarks, it leads gpt-6 luna and claude haiku 5.5 in eight of 12 tests, with significant leads in instruction following and autonomous coding. mercury 3 is designed for real-time apps like ai agents, smart glasses, and customer support, where speed matters. it can process up to 1 million tokens at once and uses an openai-compatible api. inception sees mercury 3 as a sign that diffusion models reliably improve as they scale.