Moonshot says the new model has a 1M-token context window; its faster version uses about twice the quota, and thinking cannot be disabled.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Primary source: [Kimi](https://www.kimi.com/code/docs/en/kimi-code/models.html)
Why it matters #
Kimi is extending its long-context strategy into developer workflows, where model speed, session handling and quota consumption directly affect whether a coding agent is practical to use.
Kimi says K3.1 Pro is fully rolled out in Kimi Code, which is available through desktop, command-line and editor clients. The model documentation lists a 1M-token context window for eligible plans and a faster version, K3.1 Pro Highspeed, that Moonshot says generates output about two to three times faster while using roughly twice the quota.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.