zhineng.dev
English
Get started
← All models

Kimi K3.1 Pro

Published Publisher Moonshot AIkimi-k3.1-pro

Overview

Kimi K3.1 Pro is the first model in the K3.1 series and an open-weight, multimodal model built by post-training Kimi K3. Moonshot positions it for complex coding, visual understanding, research, and long-horizon agentic tasks. Its mixture-of-experts architecture has 2.8 trillion total parameters, 104 billion activated parameters, and a 1,048,576-token context window. (kimi.ai)

Capabilities and uses

  • Supports text and image input, including native visual understanding.
  • Can assist with software development, codebase changes, tool-assisted workflows, 3D modeling and design, financial research, and financial modeling.
  • Is available through Kimi products and the Kimi API. The API also offers a separately identified Highspeed variant; its pricing is not the standard model's pricing. (kimi.ai)

Limitations

  • Outputs can contain errors or hallucinations; verify important claims, generated code, and financial analyses independently.
  • The context window is limited to 1,048,576 tokens. Long contexts can increase latency and cost.
  • According to Kimi's pricing guidance, reasoning is always enabled; reasoning_effort can be set to low, high, or max. Reported output usage, including reasoning tokens, affects cost.
  • Model weights are subject to the Kimi K3.1 Pro License; review its terms before deployment. (kimi.ai)

API pricing

The following pay-as-you-go rates apply to the standard kimi-k3.1-pro model, per 1 million tokens, excluding applicable taxes. Domestic CNY and international USD prices are separate regional offerings, not currency conversions. Checked November 15, 2026.

Platform Uncached input Output Cache hit Cache write: 5-minute TTL Cache write: 1-hour TTL
Mainland China API CNY 13.00 CNY 68.00 CNY 1.30 CNY 13.00 CNY 26.00
International API USD 2.00 USD 10.00 USD 0.20 USD 2.00 USD 4.00

For Chat Completions, uncached input, cache hits, and cache writes are mutually exclusive input billing categories. Eligible automatic caching defaults to a five-minute TTL; one hour is also available. These are standard model rates, not Highspeed, batch, or subscription prices. (kimi.com)

Official pricing sources: mainland China API guide and international API guide.

Release and official sources