From the source
Moonshot AI has released Kimi K3.1 Pro, a post-trained version of Kimi K3 with more post-training compute, which scores 55 points on the Artificial Analysis Intelligence Index and lands in fourth place.
- The model sits 3 points behind the top proprietary models, with Claude Opus 5.5 at 58, Claude Sonnet 5.5 at 56 and Muse Wonder Full at 56 ranking higher, according to Artificial Analysis.
- Moonshot is also shipping K3.1 Pro Highspeed, a variant it says runs two to three times faster, and both are available now through Kimi.ai, Kimi Work, Kimi Code and the Kimi API.
- The model keeps K3's 2.8 trillion total parameters and 104 billion active ones, spans a one million token context window, and handles text and images.
- Token prices drop by 33 percent: input costs $2 per million tokens, down from $3, output costs $10, down from $15, and cache hits cost $0.20, down from $0.30.
- Moonshot says it stripped cyber-related data from both training phases for the open-weights release, cutting its internal xdev-bench-hard score from 90.5 percent to 81.0 percent, and plans to release the model weights by November 4, 2026.