Skip to main content
AI Choice Engine

Moonshot AI

Kimi K3.1 Pro

Kimi K3.1 Pro is a Moonshot AI model with 1.05M context, $2 in / $10 out per million tokens, last verified October 25, 2026. Open weights: not verified. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.

Moonshot's new K3.1-series API model for coding and long-running agent work; the publisher describes it as 3T-class, with exact parameter count and current weight licence unverified.

Recorded profile verification October 25, 2026Individual provider fields may change

AI model specification details
SpecificationKimi K3.1 Pro
ProviderMoonshot AI
TierFrontier
Context window1.05M
Max output1.05M
Input / 1M tokens$2
Output / 1M tokens$10
WeightsNot verifiedUnverified
ParametersNot verifiedUnverified
Reasoning levelslow, high, max
Modalitiestext, image, video
API model idkimi-k3.1-pro
ReleasedOctober 23, 2026

Pricing tiers: Kimi Platform API USD/MTok input/output $2/$10, cached input $0.20; automatic 5-minute cache writes $2 and 1-hour writes $4. Separate API kimi-k3.1-pro-highspeed $4/$20, cached input $0.20 (same as regular), writes $4/$8. Taxes excluded; no context-length pricing tiers are documented. Minimum successful top-up $1. max_completion_tokens defaults 131,072 and accepts up to 1,048,576. Code memberships use different IDs, plans and quotas: regular Code model Plus/Moderato+, 1M context Pro/Allegretto+; Code Highspeed Pro/Allegretto+ and about twice regular quota. These are not API token multipliers.

Best for

  • Evaluating the current Kimi API
  • Long-context coding and agent workflows
  • Comparing standard and Highspeed API costs

Watch out

Platform API thinking is always on: low/high/max, default MAX; Code defaults HIGH. temperature=1.0/top_p=0.95/n=1/penalties=0 are fixed. Preserve complete assistant reasoning/tool history. Images use base64 or ms://, not public URLs. Web search is being updated and discouraged for near-term production use. Published API output settings do not establish a native weight-generation cap or simultaneous full input/output capacity. The API guide contains unfilled benchmark placeholders; no numeric result, K3 score or Code throughput claim is inherited. Full weight release is a future promise, not current downloadable-weight proof.

History

Kimi K3.1 Pro timeline

Release and status events for this model that passed two-source verification — dated, sourced, and linked.

  • Kimi K3.1 ProAvailability updateOctober 23, 2026

    Kimi Code announces full rollout of K3.1 Pro and its Highspeed serving variant.

    Kimi Code dated release history

Source receipts

Catalog figures for this model were checked against the following sources.

Common questions

Kimi K3.1 Pro

Answered from the verified figures on this page rather than general guidance.

How much does Kimi K3.1 Pro cost per million tokens?
Kimi K3.1 Pro lists base rates of $2 per million input tokens and $10 per million output tokens. Estimate both input and output token volume and any additional charges for your workload. Kimi Platform API USD/MTok input/output $2/$10, cached input $0.20; automatic 5-minute cache writes $2 and 1-hour writes $4. Separate API kimi-k3.1-pro-highspeed $4/$20, cached input $0.20 (same as regular), writes $4/$8. Taxes excluded; no context-length pricing tiers are documented. Minimum successful top-up $1. max_completion_tokens defaults 131,072 and accepts up to 1,048,576. Code memberships use different IDs, plans and quotas: regular Code model Plus/Moderato+, 1M context Pro/Allegretto+; Code Highspeed Pro/Allegretto+ and about twice regular quota. These are not API token multipliers.
What is Kimi K3.1 Pro's context window?
Kimi K3.1 Pro accepts about 1.05M tokens of context. That only matters if you routinely send very long documents, large codebases, or multi-turn histories that approach that limit.
What is Kimi K3.1 Pro best for?
Kimi K3.1 Pro is a frontier tier from Moonshot AI. It suits evaluating the current kimi api, long-context coding and agent workflows, comparing standard and highspeed api costs. Platform API thinking is always on: low/high/max, default MAX; Code defaults HIGH. temperature=1.0/top_p=0.95/n=1/penalties=0 are fixed. Preserve complete assistant reasoning/tool history. Images use base64 or ms://, not public URLs. Web search is being updated and discouraged for near-term production use. Published API output settings do not establish a native weight-generation cap or simultaneous full input/output capacity. The API guide contains unfilled benchmark placeholders; no numeric result, K3 score or Code throughput claim is inherited. Full weight release is a future promise, not current downloadable-weight proof.
Model comparison catalogBenchmark library