Unverified50% confidenceFactExact time
DeepSeek V4.1 Flash (version number labeled 10.17) runs at 70-80 tokens/second on Matt's local DGX Station, with power consumption of about 1600 watts.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/23/2026
First Seen
Valid until: 11/6/2026
Sources
Kimi K3.1 Hands-On Test: Open-Source Model Nears Opus, Video Generation Steals the Show
youtubeMatthew Berman10/23/2026
Related Entities
Related Claims
UnverifiedThe closed-source Hologen Flash Server (Linux) can achieve a prefill speed of 1,200 to 1,400 tokens, outperforming GFX1151 Engine.79% similarUnverifiedThe RX 7900 XTX can achieve speeds of 60–100 tokens/second when running Flash-series models of approximately the 7B scale.73% similarUnverifiedDeepSeek V4 Flash is a cloud-based model focused on high throughput, with an inference speed of approximately 140 tokens per second.73% similarUnverifiedThe DeepSeek Flash model can output more than 100 tokens per second.72% similarUnverifiedZhipu has released GLM-4.5 FlashX, reaching a maximum speed of 200 tokens/second, with the API now live.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1091151API
curl https://kongchang.com/api/v1/knowledge/claims/1091151MCP
get_claim(id=1091151)