Dev Radar
Support
LiveUpdated 2026-11-02 18:17 UTC

DeepSeek-V4.1-Flash-1017 NVFP4 on 4 DGX Sparks: @MiaAI_lab's TensorFold recipe decodes…

DeepSeek-V4.1-Flash-1017 NVFP4 on 4 DGX Sparks: @MiaAI_lab's TensorFold recipe decodes code 2.1x faster than vLLM. 4…

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

DeepSeek-V4.1-Flash-1017 NVFP4 on 4 DGX Sparks: @MiaAI_lab's TensorFold recipe decodes code 2.1x faster than vLLM. 4 Sparks, Mia's recipe: Decode, code: 71.0 tok/s Decode, prose: 68.4 tok/s Decode, HTML answer: 99.8 tok/s 500 words: about 10.3 s First token, 8K: 2.2 s 1 at once: 62.1 tok/s 8 at once: 144.6 tok/s 4 Sparks, vLLM: Decode, code: 33.6 tok/s Decode, prose: 30.2 tok/s Decode, HTML answer: 63.3 tok/s 500 words: about 22.4 s First token, 8K: 2.2 s 1 at once: 27.6 tok/s 8 at once: 114.8 tok/s The decode gap narrows with load, 2.3x at 1 at once and 1.3x at 8. Prefill is close: vLLM is

Posted by Niklas (2.7k followers) 3 h ago · 22 likes · 1.8k views · view the original post on X. Kept by the Dev Radar as AI dev tools. Tools mentioned: niklaslenz.com.

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. 31.3k posts from 5k X accounts over the last 21 days, 9k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-02 18:17 UTC. Full method.