Inception's diffusion LLM Mercury 2 named TIME Best Invention, 1000+ tok/s
volokuleshov · x · 2026-10-16
Inception's diffusion-based LLM Mercury 2 made TIME's Best Inventions of 2026. Unlike autoregressive models, it generates a noisy draft then refines entire text blocks in parallel, reaching 1000+ tokens per second in production. CEO Stefano Ermon claims it's 5–10x faster than models behind consumer Claude and ChatGPT — a striking turnaround for what was a niche discrete-diffusion research bet.
Related event: Diffusion LLM Mercury 2 Named TIME Best Invention, Bigger Model Coming(4 posts)→
More from Infra
- Berkeley scholars: concentrated chip supply chain makes a hardwired frontier AI pause feasible — wfithian · 2026-10-16
- Microsoft trained MAI-Base-2 on 16k GB200 GPUs in just 38 days, with 21% MFU — scaling01 · 2026-10-16
- Robotics data tokenization landscape: Khala maps peaq, THOR, BitRobot and Axis — 0xSammy · 2026-10-16
- Signal65 benchmarks: Intel Xeon 6 runs agent workloads 11% faster for less money — ryanshrout · 2026-10-16
- Microsoft's MAI-Base-2: 2.8T-param MoE pretrained on 16k GB200s in just 38 days — scaling01 · 2026-10-16
- Cloudflare and OpenAI pilot: real-time network signals bring unseen pages into ChatGPT search — irvinebroque · 2026-10-16