Kimi K3.1 Pro

Post-trained Kimi K3 for coding and agent work, with a 1M-token context.

Kimi K3.1 Pro is Moonshot AI’s post-trained successor to Kimi K3, aimed at coding, financial research and other long agent tasks. It reads text, images and video, always reasons before answering, and costs less per token than K3. It is on the Kimi API and apps now; Moonshot says the full weights will be published by November 4, 2026.

Provider: Moonshot AI
Type: Language model
Released: Oct 23, 2026
Context window: 1M tokens
Price: $2.00 input / $10.00 output per 1M tokens
Input: text, image, video
Output: text
Open weights: No

Best for

Coding
Agents
Reasoning
Research
Long context

Strengths

1M-token context
Video input
Cheaper than Kimi K3

More from Moonshot AI

Kimi K3 — Moonshot AI. Moonshot AI’s 2.8T-parameter open multimodal reasoning model.
Kimi K2.7 Code — Moonshot AI. Coding-focused Kimi K2 model for long, end-to-end programming tasks.

Sources