Kimi K3.1 Pro

Moonshot AI · Released Oct 23, 2026

Input price / MTok

$10

Output price / MTok

1.0M

Context window

73.5

Tokens / sec

65.7

Terminal-Bench 4.0

Artificial Analysis Intelligence Index

Summary

Kimi K3.1 Pro is a proprietary language model from Moonshot AI, released on Oct 23, 2026.

At $2 input and $10 output per million tokens, it ranks #27 of 40 for input price, placing it mid-range on input cost.

Its context window of 1.0M tokens ranks #26 of 41 for context depth in this dataset.

It targets coding, agents, reasoning workloads.

Throughput is measured at 73.5 tokens per second (ranked #16 of 23 for throughput).

Post-trained version of Kimi K3 (2.8T total / 104B active MoE). Always-on thinking with reasoning_effort low/high/max (default max). Text, image and video input. Moonshot says full weights ship by Nov 4 2026; until then it is API-only, so openSource is false. Cyber-related training data is removed from the open-weight release.

Benchmarks

Technical specs

Moonshot AI

Similar Models

Models with the most shared benchmark coverage and closest scores to Kimi K3.1 Pro.

Compare Kimi K3.1 Pro side by side

Pre-filled with 3 similar models

Speed source: https://artificialanalysis.ai/models/kimi-k3-1-pro