The model is a mixture-of-experts system with 2.8 trillion total parameters, 104 billion active parameters, and a stated context window of up to one million tokens.
The standard API price is $2.00 per million input tokens and $10.00 per million output tokens. Moonshot also offers a Highspeed version described as two to three times faster, priced at twice the standard rate except for input cache hits. The model supports low, medium, and max reasoning settings.
Kimi K3.1 Pro is not yet available as open weights. Moonshot says it will release the full model weights by November 4, 2026. The supplied source reports benchmarks published by Moonshot and an Artificial Analysis score, but those results do not establish independent performance for a particular workload, deployment, latency target, or cost profile.