Moonshot AI releases Kimi K3.1 Pro with 1 million token context window

On October 23, 2026, Beijing-based AI startup Moonshot AI launched Kimi K3.1 Pro, featuring a 1-million-token context window and improved capabilities in coding, finance, and 3D modeling. Valued at approximately $50 billion ahead of a planned Hong Kong stock debut in early 2027, Moonshot AI reduced API pricing compared to K3 while planning an open-weight release for November 2026. The release comes amid allegations of distillation from U.S. competitors and new Chinese export registration draft rules.

Kimi K3.1 Pro model release

• ▪On October 23, 2026, Beijing-based Moonshot AI released Kimi K3.1 Pro, the first model in its K3.1 series, featuring a 1-million-token context window
• ▪Moonshot AI plans to release open weights for the Kimi K3.1 Pro model in November 2026
• ▪Moonshot AI released Kimi K3.1 Pro Highspeed alongside Kimi K3.1 Pro, which runs two to three times faster according to official company statements

Technical architecture specifications

• ▪Kimi K3.1 Pro builds on the base K3 architecture utilizing Kimi Delta Attention and Attention Residuals, trained with joint reinforcement learning across hundreds of downstream domains
• ▪Moonshot AI removed offensive cyber-related data from mid-training and reinforcement learning for Kimi K3.1 Pro, lowering its pass@3 score on xdev-bench-hard from 90.5% in an unfiltered checkpoint of Kimi K3.1 Pro to 81.0%
• ▪Kimi K3.1 Pro is a 2.8-trillion total parameter Mixture-of-Experts model that activates 104 billion parameters across 16 of 896 experts using a Stable LatentMoE framework

Benchmark performance comparison

• ▪On the internal Kimi Hallu Bench Hard stress test, claim-level hallucinations for Kimi K3.1 Pro fell to 8.3% from 23.4% in K3, compared to 5.5% for GPT-6 Astra
• ▪On financial benchmarks, Kimi K3.1 Pro reported a score of 64.4 on OfficeQA Pro v2 and 58.6 on Frontier Finance
• ▪Kimi K3.1 Pro scored 65.7% on Terminal-Bench 4.0 coding, trailing Claude Opus 5.5 at 66.4%, while scoring 38.1 on Terminal-Bench Science 0.1 compared to 68.1 for OpenAI's GPT-6 Astra
• ▪On the SWE-Marathon coding benchmark, Kimi K3.1 Pro scored a self-reported 51.7, leading Anthropic's Claude Opus 5.5 score of 46.3

API pricing structure

• ▪Kimi K3.1 Pro Highspeed API pricing is set at $4.00 per million input tokens and $20.00 per million output tokens, compared to Kimi K3's rates of $3.00 input and $15.00 output
• ▪Moonshot AI set API pricing for Kimi K3.1 Pro at $2.00 per million input tokens and $10.00 per million output tokens, with cache hits costing $0.20 per million tokens

Developer access methods

• ▪Kimi K3.1 Pro is accessible via Kimi.ai, Kimi Work, Kimi Code, and an OpenAI-compatible API using API IDs kimi-k3.1-pro and kimi-k3.1-pro-highspeed
• ▪Kimi K3.1 Pro API requires a minimum $1 account top-up to unlock access, defaults output to 131,072 tokens with a maximum of 1,048,576, and supports vision input via base64 or uploaded file IDs

Moonshot AI competitive positioning

• ▪Chinese regulators proposed draft export rules requiring AI companies to register before overseas deployment, under which DeepSeek and MiniMax Group Inc. have already secured approvals
• ▪Moonshot AI's annual recurring revenue reached over $1 billion in August 2026, up from $300 million in June 2026, with a target of reaching $2 billion in annualized sales by the end of 2026
• ▪Moonshot AI closed a private fundraising round at a valuation of approximately $50 billion ahead of an anticipated Hong Kong initial public offering in the first quarter of 2027