Inception Labs has launched Mercury 3, a diffusion-based large language model that claims to outperform models like GPT-6 Luna and Claude Haiku 5.5 in speed and quality. The model boasts a 1 million token context window and can process over 1,000 tokens per second on NVIDIA GPUs, making it suitable for latency-sensitive applications. Mercury 3 is positioned as a scalable diffusion model capable of competing with autoregressive models on complex tasks, with early access partners already utilizing it for agents, education, search, and customer support. AI
IMPACT Sets new SOTA for diffusion LLMs, potentially accelerating adoption in real-time applications.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Hacker News — AI stories ≥50 points →
- Claude Haiku 5.5
- gpt-6
- GPT-6 Luna
- Mercury 2.5
- Mercury I
- Mercury III
- Mind2Web
- Nvidia
- Shuhao Zhang
- TinyFish AI
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →