Dev Radar
Support
LiveUpdated 2026-10-31 14:16 UTC

DeepSeek-V4.1 Flash 1017 update on two DGX Sparks.

DeepSeek-V4.1 Flash 1017 update on two DGX Sparks. Updated to the latest @tensorfold_dev, thanks to me @ashxhart and…

This is a dev post classified by AGK Intelligence as AI dev tools (a launch), kept by the Dev Radar because it carries real work, not commentary.

DeepSeek-V4.1 Flash 1017 update on two DGX Sparks. Updated to the latest @tensorfold_dev, thanks to me @ashxhart and the rest of the TensorFold community. 100.9 tok/s on a real coding task with MTP. 137.4 tok/s across 4 streams. Prefill 2,203 tok/s at 128K. The rest of the performance is in the draft acceptance, it failed a few tests if I turned that up any more with my quant. The new build is almost the same speed as before, with identical output and strict tool calling passing. I adjusted the benchmarks to follow a bit closer to what everyone else is doing to bring things into line.

Posted by JJ Eaton (1.5k followers) 1 h ago · 21 likes · 1.1k views · view the original post on X. Kept by the Dev Radar as AI dev tools.

More dev work like this

Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 31.2k posts from 5k X accounts over the last 21 days, 8.9k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-10-31 14:16 UTC. Full method.