Kimi tops open-weight intelligence index at 46, but 2.8T-param model is the priciest to run
drdanielbender · x · 2026-11-13
A practical shortlist of 4 open-weight models, spanning 27B (Qwen) up to 2.8T total / 104B active parameters.
- Kimi K3.1 Pro leads the Artificial Analysis intelligence index among open-weight models at 46 points, and is the biggest model on the list — but it's also the most expensive option for both self-hosting and API use.
- The author's point: the most interesting AI model isn't always the smartest one — regional compliance and running on GPUs you already own often matter more.
- Full details of all four picks are in the thread.
Related event: Four Open-Weight Models Worth Running Locally, from 27B to 2.8T(4 posts)→
More from Models
- Every's Vibe Check: Fable 5.2 Runs Unattended Overnight and Built a Video Editor in Days — every · 2026-11-13
- Sakana AI Starts Pretraining Its Own Model from Scratch, Weights Due Spring 2027 — tkasasagi · 2026-11-13
- Bug Hunt Bench leaderboard ranks coding models on planted-bug fixes, cost and runtime — PawelHuryn · 2026-11-13
- Redditor pits Gemini 4 Hydrogen against a zero-asset 3D site prompt: 'Opus 5.5 territory' — JustA_Curious_Human · 2026-11-13
- VC's hard-won take: LLM advice is fluent but often wrong — you must push back to get value — StewartalsopIII · 2026-11-13
- Claude Pro's 'endless tokens' reportedly ending this December, user warns — D3VAUX · 2026-11-13