Kimi K3.1 Pro: 2.8T-parameter 'open-weight' model you still can't download
alex_verem · x · 2026-10-29
Moonshot AI announced Kimi K3.1 Pro on Oct 23 as "our most capable 3T-class open-weight model": 2.8T total parameters, 104B active, 1M-token context, image+text input. But as alexverem found, the promised weights (due by Nov 4) hadn't landed on Hugging Face as of Oct 29 — the latest listed model is still Kimi K3, and K3.1 Pro is only accessible via Moonshot's apps and paid API ($2/$10 per million tokens in/out).
Key numbers from the accompanying charts:
- Kimi K3 file sizes: 466 GB (Unsloth 1-bit, needs 610 GB memory), 594 GB recommended, 861 GB 2-bit (one 8×B200 server), 1.56 TB full quality
- Four ways to use K3.1 Pro: official app/API (available now), third-party hosts, rented 8×B200 server ($53.52/hour), or your own 610 GB+ workstation
Bottom line: no home computer will run the full model; most businesses will opt for third-party hosting once weights ship.
More from Infra
- Oracle plans 10GW capacity by FY28 as AI backlog jumps 4.6x to $638B in one year — rohanpaul_ai · 2026-10-29
- Unsloth quants run 1T-param Mistral Large 4 on Mac Studio: 1-bit keeps 91% accuracy at 223GB — danielhanchen · 2026-10-29
- SMCI-backed neocloud Argentum AI seeks $5B credit facility for chip purchases — Beth_Kindig · 2026-10-29
- Microsoft returns to bond market seeking $25B+; AI debt wave lifts borrowing costs for all — rohanpaul_ai · 2026-10-29
- Self-hosted realtime avatar server: 30 FPS to 90 FPS on a free Colab T4 — SavingsWeather1659 · 2026-10-29
- WSJ: Apple refused flash deliveries when prices fell, leaving suppliers stuck — firstadopter · 2026-10-29