Same chat, same model: 52 tok/s writing code, 9 tok/s explaining something. Timed start…
This is a AI post classified by AGK Intelligence as AI infra & evals (a model release), kept by the AI Radar because it carries real work, not commentary.
Same chat, same model: 52 tok/s writing code, 9 tok/s explaining something. Timed start to finish. Turns out the small "draft" model that guesses ahead is way better at guessing code. Ricursive told me they trained it mostly on code, so that tracks. I've tried a bunch of tweaks since. Some prompts got faster, some got slower. Nothing worth keeping. Next up: plain vLLM on one Spark, then split across both. Same model, 128K context. I just want a single chat to feel fast. Anyone actually getting this to speed up regular writing?
Posted by Joey (5k followers) 11 h ago · 4 likes · 336 views · view the original post on X. Kept by the AI Radar as AI infra & evals.
More AI work like this
- websites bay area people use that normies have never heard of: — @chiefofautism
- Coding agents can ship an app in hours. — @DataChaz
- Introducing TRACE: Transparent Real-world Assessment of Content Extraction. — @getcontextdev
- We're sitting down with and @cjav_dev (@AnthropicAI) on Nov 17th at 10am PT to talk… — @sentry
- Introducing our next generation of model routers: — @OpenRouter
- a16z's Martin Casado: AI is the biggest rebuild of computer science since the internet.… — @guilleflorvs
- There are AI GPUs running in converted chicken coops. — @guilleflorvs
- Enterprise AI is only as useful as the data and business context behind it. — @databricks
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. 65.3k posts from 5k X accounts over the last 14 days, 10.4k tools, 19 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 17:44 UTC. Full method.