Kimi researcher says U.S. frontier labs can tap thousands of GPUs
dotey · x · 2026-07-21
A Kimi researcher says frontier labs in the U.S. have access to thousands of GPUs, and that this is genuinely shocking for researchers coming back to China.
The post frames compute scarcity as a major obstacle to frontier innovation in China, while also citing an informal model ranking that places Fable 5 first, GPT-5.6 Sol second, Kimi K3 third, Grok 4.5 fourth, and GLM 5.2 fifth. It adds a personal view that GLM could improve substantially with much larger parameter scales.
Related event: Kimi Researcher: GPU Scarcity Hinders China's AI Innovation(3 posts)→
More from Infra
- Spomin: live KV cache compaction squeezes 500k tokens of context into 180k resident — wgaca2 · 2026-09-11
- PiPNN nearest-neighbor search wins three awards, up to 78x faster index building — khademinori · 2026-09-11
- M.2-Oculink eGPU Link Silently Downgrades to PCIe Gen1 — Here's How to Check — El_90 · 2026-09-11
- DeepSeek launches V4.1-Flash with 1M-token context and 4x smaller KV-cache — matlabulous · 2026-09-11
- What Can You Still Run on 8GB VRAM? User Asks for Small Models With Tool Use — riceinmybelly · 2026-09-11
- Spain's hourly 80% renewable matching rules clash as France fast-tracks 700MW sites, UK cuts grid queues — eherrerosj · 2026-09-11