DeepSeek-V4.1-Flash-1017: a Mac Studio decodes code at 80.5 tok/s, 4 DGX Sparks get the…
This is a dev post classified by AGK Intelligence as Hosting & infra (a tool drop), kept by the Dev Radar because it carries real work, not commentary.
DeepSeek-V4.1-Flash-1017: a Mac Studio decodes code at 80.5 tok/s, 4 DGX Sparks get the first token 2.1x sooner at 32K 4 Sparks (512 GB), Mia's recipe: Decode, code: 69.5 tok/s Decode, prose: 70.2 tok/s 500 words: about 10.0 s First token, 32K: 8.8 s 1 at once: 60.5 tok/s 4 at once: 115.9 tok/s 8 at once: 139.0 tok/s Mac Studio (256 GB), Bob's recipe: Decode, code: 80.5 tok/s Decode, prose: 70.5 tok/s 500 words: about 9.3 s First token, 32K: 18.6 s 1 at once: 62.5 tok/s 4 at once: 88.2 tok/s 8 at once: 88.3 tok/s One model, two boxes, each on its maker's recipe and settings. Speed v2: same
Posted by Niklas (2.3k followers) 7 h ago · 22 likes · 2.7k views · view the original post on X. Kept by the Dev Radar as Hosting & infra.
More dev work like this
- I like that our @googlecloud spend caps (for Gemini API, Cloud Run, Gemini platform) now… — @rseroter
- Sleeper GPU of the Week 💤 — @vast_ai
- The era of personal software is so wild — @allietheicon
- Your first Kubernetes cluster and your tenth should install and upgrade the same way. — @CtrlIQ
- #developers, Love using #Slack to deploy and manage your apps? Now chat with #Slackbot… — @scottcjohnston
- Using Akash Console with other people used to mean sharing one login. No more. — @akashnet
- Regenerating database passwords or bucket credentials in the dashboard now automatically… — @Railway
- Este repo tiene miles de ejercicios para DevOps, arquitectura y servidores. — @midudev
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, 8.8k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-10-30 20:27 UTC. Full method.