Moonshot launches Kimi K3.1 Pro; weights promised November 4
Original title: [AINews] Kimi 3.1 Pro 104BA-2.8T: Fable-5.1-Class Open Weights narrows Closed-AI lead to 1 month... and Pro Highspeed API with 2-3× speed
The 2.8T mixture-of-experts model, with 104B active parameters, scored 55 on Artificial Analysis; the API is live but weights are not.
Moonshot launched Kimi K3.1 Pro, described as a post-trained multimodal mixture-of-experts model with 2.8 trillion total parameters, 104 billion active parameters, and a 1-million-token context window. Standard and Highspeed API access is available; Moonshot says Highspeed is 2–3× faster and costs twice as much. Weights are promised by November 4, 2026, and were unpublished at launch, so license and inference requirements are not established.
Unconfirmed: Weights, license, and serving requirements are unpublished; some capability and speed figures are Moonshot's claims.
- Artificial Analysis ranked it fourth on its Intelligence Index, scoring 55.
- Arena placed it eighth in Code Arena: WebDev at 1,692 points.
- It scored 64% on Artificial Analysis' Cyber Index, second overall in that listing.
- Kimi Hosted Agents adds a managed environment for long-running multi-agent work.
- Vendors
- Moonshot
- Products
- Kimi Hosted Agents
- Organizations
- Artificial Analysis
- AI models
- Kimi K3.1 ProKimi K3
From the original article
12 min read at latent.space
We have Fable at home... or so we are being told by paid shills
By swyx .ai · latent.space
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.