- Published on
Introducing Kimi K3.1 Pro
- Authors
- Name
- DeepHorus
Moonshot has released Kimi K3.1 Pro on the Kimi API, making its newest flagship model available to developers on 23rd October 2026. The model is not yet open weights, but the company has promised that full model weights will be released by 4th November 2026. The launch places Kimi K3.1 Pro squarely in the top tier of frontier models while retaining a commitment to eventual open release.
Kimi K3.1 Pro follows the earlier Kimi K3 and is positioned as Moonshot's most capable model to date. According to Moonshot's own benchmarks, the new release is approximately level with Claude Opus 5.5 and GPT-6 Astra, beating both on a handful of benchmarks and coming close on others. It marks the first time a soon-to-be open weight model has scored this highly against the frontier.
Pricing is set at $2.00 per million input tokens and $10.00 per million output tokens. That matches GPT-6.1 Sol and is half the price of Claude Opus 5.5. Moonshot is also offering a Highspeed variant that runs two to three times faster at twice the price, although input token cache hits are priced the same as on the standard model.
Independent evaluator Artificial Analysis scored Kimi K3.1 Pro at 55. That places it ahead of Claude Fable 5.1, GPT-6 Astra and Gemini 4 Argon, which all scored 53, and behind only Claude Opus 5.5 at 58. The gap to the leading proprietary model is now just four points.
Early hands-on testing ran the model through the Moonshot API using the llm openai endpoint command added to LLM 0.32 in August. The test generated an SVG of a pelican riding a bicycle at each of the three reasoning levels the model supports — low, medium and max — with fresh comparisons run against Kimi K3 at the same settings. The exercise showed how easily existing OpenAI-compatible tooling can be pointed at the new endpoint.
The mix of frontier-adjacent scores and aggressive pricing increases pressure on rival labs. At half the cost of Claude Opus 5.5, Kimi K3.1 Pro offers developers a cheaper option for demanding workloads. The promise of open weights, meanwhile, would let organisations self-host a model whose capabilities until recently would have required a proprietary API.
Attention now turns to 4th November 2026, when Moonshot has said the full weights will be published. If that deadline holds, Kimi K3.1 Pro would become the strongest openly available model to date. Competitors are likely to respond on both price and capability in the weeks that follow.
