All release briefs
UpdatedFull rollout announced

Moonshot AI / Kimi K2.8 Preview

Kimi Code gets a new model behind the same name.

K2.8 Preview replaces the model behind kimi-for-coding, with adjustable reasoning and a larger context allowance.

LanguageKimi Code

SCOUT’S TAKE

Existing integrations receive the update without choosing a new model identifier.

Access
Kimi Code membership
Model identifier
kimi-for-coding
Context
Up to 1M tokens
Reasoning settings
Low / high / max

What changed.

  1. Moonshot announced the full Kimi Code rollout on September 11.

  2. Reasoning effort offers low, high and max settings.

  3. All membership tiers can use up to one million context tokens.

Kimi Code official changelog

Before you use it.

Check your client’s context setting and membership limits. This announcement does not establish downloadable weights or a standalone token price.

What the evidence shows

Moonshot reports improved coding and reasoning efficiency. Scout has not independently measured those improvements. The date refers to the announced full rollout.

Read another briefSWE-2