Models · Updated 19 Oct, 10:57 pm IST
Mercury 3: diffusion LLM claims frontier performance and 1,068 tok/s
Why it matters for readers: If you follow AI models, this claim suggests diffusion-based approaches may now compete with top autoregressive models on speed and capability.
- Mercury 3 is presented as the most capable diffusion-based LLM to date and reportedly outperforms GPT-6 Luna and Claude Haiku 5.5 on several benchmarks.1
- The team reports Mercury 3 runs at 1,068 tokens per second on NVIDIA GPUs and is 4–8× faster in high-throughput scenarios.1
- Mercury 3 increases context length to 1 million tokens (up from 260K in Mercury 2.5) and is over 40× larger than Mercury 1.1
- Pricing at launch is $0.20 per million input and $0.70 per million output tokens, with a 50% introductory discount and deeper discounts for cached tokens.1
- Early-access partners are using Mercury 3 for voice and chat support, real-time tutoring, data engineering, and web navigation according to the announcement.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get started