Moonshot prices the model at $2 per million input tokens and $10 per million output tokens, level with GPT-6.1 Sol.

Moonshot launched Kimi K3.1 Pro on the Kimi API on October 23, 2026, with weights not yet open but promised by November 4, 2026. The company also sells a Highspeed tier that runs two to three times faster at twice the price, though cached input tokens cost the same in both tiers. Pricing lands at $2.00 per million input tokens and $10.00 per million output tokens.

Moonshot’s own benchmarks place Kimi K3.1 Pro roughly level with Claude Opus 5.5 and GPT-6 Astra, winning a few tests and nearing them on others. Artificial Analysis scored the model at 55, ahead of Claude Fable 5.1, GPT-6 Astra, and Gemini 4 Argon at 53, and behind only Claude Opus 5.5 at 58. The result marks the first time a soon-to-be open weight model has reached this level against frontier systems.

Builders weighing Kimi K3.1 Pro against closed frontier models gain a cheaper option with comparable benchmark scores and a published price list. The Highspeed tier suits latency-sensitive agent loops, while cached input pricing keeps repeated system prompts affordable. Teams that need weights on their own infrastructure must wait for the November release before they can self-host.

The next checkpoint is November 4, 2026, when Moonshot has promised to publish full weights. Until then, the Kimi API is the only way to run the model, and the Highspeed tier gives buyers a faster path at double the token rate. Watch whether independent evaluators reproduce the 55 score once weights ship and the community can run its own tests.

What matters

• Moonshot says full model weights for Kimi K3.1 Pro will arrive by November 4, 2026.
• Builders get a near-frontier model at half the price of Claude Opus 5.5 for agent workloads.
• Watch whether Moonshot ships weights on schedule and how the Highspeed tier prices at scale.

Why it matters

Watch whether Moonshot ships weights on schedule and how the Highspeed tier prices at scale.

This GenAI News article was prepared in original wording using reporting and materials published by Simon Willison’s Weblog. Source reference: https://simonwillison.net/2026/Oct/23/kimi-k31-pro/.

Drafted by the GenAI News review pipeline.