Oriel — AI & technology worth noticing, every day.

Kimi K3.1 Pro and K3.1 Pro Highspeed

Web page · https://kimi.com/code/docs/en/kimi-code/models.html

First seen by Oriel on .

AI summary

Kimi Code documents six model IDs, their capabilities and plan requirements, and how to select them in official clients or configure third-party API tools.

  • K3.1 Pro offers a 1M context window and thinking-only operation; K3.1 Pro Highspeed has the same capabilities at approximately 2–3× speed and uses about twice the quota.
  • The lineup also includes K3, K3-256k, kimi-for-coding, and kimi-for-coding-highspeed, with different context limits, speeds, and plan access.
  • Switching models or thinking effort invalidates the context cache and increases token usage, so the guide recommends keeping both consistent within a session.
  • Third-party tools can use OpenAI Responses or Chat Completions and Anthropic-compatible protocols; requests should specify the exact model ID and the appropriate regional Base URL.
  • Invalid model IDs may silently fall back to K2.8 Preview without thinking, while unsupported permissions, thinking settings, or context lengths can produce HTTP 400 or 403 errors.

#kimi code #llm models #api configuration #context windows

Generated by gpt-5.6-sol. Check the original for details.

Metric trends

Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained

Points

Score reported by the source.

  • Hacker News: 1,088, latest observation ; +1,010 over observed period since

Comments

Comment count reported by the source.

  • Hacker News: 674, latest observation ; +658 over observed period since

Aliases

SourceExternal IDURL
Hacker News50212065https://www.kimi.com/code/docs/en/kimi-code/models.html Read discussion