Inception releases Mercury 3, a diffusion language model it says runs at 1,068 tokens per second with a 1-million-token context

Inception released Mercury 3 on 19 October. Its post opens: “Today we're introducing Mercury 3, the most capable diffusion-based LLM released to date.” Inception's models generate text by diffusion rather than word by word, which is the basis of its speed claims.

Inception gives a speed of 1,068 tokens per second on Nvidia GPUs, and says that is 4 to 8 times faster than OpenAI's GPT-6 Luna and Anthropic's Claude Haiku 5.5. It says Mercury 3 is on par with or better than both, but publishes no standard benchmark scores to show it. The one named test is its own live run of 50 Mind2Web browser tasks, in which it says Mercury 3 completed 50%.

The context window grows to 1 million tokens, from 260,000 in Mercury 2.5. The price is $0.20 per million input tokens and $0.70 per million output tokens, with 50% off at launch. The model is available through Inception's API, OpenRouter, Baseten and Vercel.

Inception made its diffusion voice model, Mercury Voice, generally available on 29 September, as the Switchboard reported. For Mercury 3's claim to match mainstream small models on quality, not only on speed, the evidence so far is Inception's own.

• Confirmed Mercury 3 has a 1-million-token context and costs $0.20 per million input and $0.70 per million output tokens. Inception
• Claimed It runs at 1,068 tokens per second, 4 to 8 times faster than GPT-6 Luna and Claude Haiku 5.5, Inception says. Inception