Kimi 3.1 Pro: 2.8T open model narrows closed-AI lead to one month
In brief
- Kimi 3.1 Pro is a multimodal 2.8T-parameter model with 104B active parameters and a 1M-token context window.
- It outperforms Fable 5.1 on some benchmarks and ranks 4th on Artificial Analysis’ Intelligence Index (score 55).
- Open weights are promised for November 4, 2026, but technical and commercial details remain unclear.
Moonshot launched Kimi 3.1 Pro, a multimodal model with 2.8 trillion parameters, 104 billion active parameters, and a one-million-token context window. According to the official announcement, it outperforms Fable 5.1 on some benchmarks and ranks fourth on Artificial Analysis’ Intelligence Index with a score of 55, behind Opus 5.5 (58) and Sonnet 5.5 (56).
The model is available in standard and Highspeed versions via Kimi’s API, website, and apps, with speed gains of 2 to 3 times for the Highspeed tier. Moonshot promises an open-weights release on November 4, 2026, but precise technical details (license, commercial terms, checkpoint precision) remain unspecified. Early feedback highlights improvements in coding, financial modeling, and collaborative agent work, alongside concerns about inference economics and benchmark quality.
This release represents a milestone for open models, narrowing the gap with closed ones according to Artificial Analysis, which estimates the gap on its index would shrink from 12 to 3 points once weights are published. Reactions range from enthusiasm over the model’s competitiveness to skepticism about its deployment economics and conditions.
This summary gives you the essentials. The full article, its figures and quotes are on Latent Space.
Read the original articleAI-generated summary of the source, checked by rules. How we write · Report an error