From the source
Inception released Mercury 3, which it calls the most capable diffusion-based LLM to date, claiming it outperforms GPT-6 Luna and Claude Haiku 5.5 while running 4–8x faster at over 1,000 tokens per second.
- The company says Mercury 3 is more than 40x larger than Mercury 1 and answers the open question of whether diffusion LLMs can scale to compete with autoregressive models on complex use cases.
- Inception reports that intelligence across standard benchmarks has risen 3x from Mercury 1 (mini) to Mercury 3, and that the scaling laws behind autoregressive progress hold for diffusion too.
- Mercury 3 carries a 1M token context window, up from 260K for Mercury 2.5, and is priced at $0.20 per million input and $0.70 per million output tokens, with a launch discount of 50% bringing it to $0.10 and $0.35.
- It is available today on the Inception API Platform and through OpenRouter,
Baseten and
Vercel, is
OpenAI API compatible, and comes with 100 million free tokens for API trials.
