Moonshot AI

Kimi K2.8 Preview

Name: Kimi K2.8 Preview
Author: Moonshot AI

Kimi K2.8 Preview is a Moonshot AI model with 1.05M context, Not verified in / Not verified out per million tokens, last verified October 2, 2026. Open weights: not verified. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.

Kimi Code’s preview model served through the existing kimi-for-coding alias, with text, image and video input and adjustable thinking.

Recorded profile verification October 2, 2026Individual provider fields may changeAccess: preview — Fully rolled out September 11 under the existing kimi-for-coding alias while still named Preview. Plus+ new plans or Andante+ legacy plans have coding access; Go has no coding quota.

AI model specification details
Specification	Kimi K2.8 Preview
Provider	Moonshot AI
Tier	Balanced
Context window	1.05M
Max output	Not verifiedUnverified
Input / 1M tokens	Not verifiedUnverified
Output / 1M tokens	Not verifiedUnverified
Weights	Not verifiedUnverified
Parameters	Not verifiedUnverified
Reasoning levels	low, high, max
Modalities	text, image, video
API model id	kimi-for-coding
Released	September 11, 2026

Pricing tiers: Kimi Code membership/quota route, not a verified pay-as-you-go per-token tariff. Current model table lists Plus and above for new plans and Andante and above for legacy plans; the Go tier has no coding quota. Do not transfer K3 platform token prices to this alias.

Best for

Evaluating Kimi Code integrations
Image and video context in coding workflows

Watch out

Still named Preview despite the September 11 full rollout. Default effort is max; native choices low/high/max. Gateway aliases ultra/xhigh map to max, medium to high, minimum/light to low, none disables thinking; unknown values return HTTP 400. Thinking-disabled K3 and K2.8 requests are served by K2.8 without thinking. Changing model or effort invalidates context cache. The exact Code model table gives 1,048,576 context and image/video input; membership/quota rules apply. Maximum output, parameters, standalone weights/licence and exact comparable benchmark scores remain unverified.

History

Kimi K2.8 Preview timeline

Release and status events for this model that passed two-source verification — dated, sourced, and linked.

Kimi K2.8 PreviewLimited previewSeptember 11, 2026
Moonshot rolls out Kimi K2.8 Preview in Kimi Code under the existing `kimi-for-coding` alias, with up to 1,048,576 context tokens; no pay-as-you-go token tariff is verified.
Kimi Code — what's new

Source receipts

Catalog figures for this model were checked against the following sources.

Current exact Code model/effort/membership matrix, alias mappings and API bases (accessed 2026-10-02)
Original dated publisher HTML: September 11 full K2.8 rollout; extracted headings omit date (accessed 2026-10-02)

Common questions

Kimi K2.8 Preview

Answered from the verified figures on this page rather than general guidance.

What is Kimi K2.8 Preview's context window?

Kimi K2.8 Preview accepts about 1.05M tokens of context. That only matters if you routinely send very long documents, large codebases, or multi-turn histories that approach that limit.

What is Kimi K2.8 Preview best for?

Kimi K2.8 Preview is a balanced tier from Moonshot AI. It suits evaluating kimi code integrations, image and video context in coding workflows. Still named Preview despite the September 11 full rollout. Default effort is max; native choices low/high/max. Gateway aliases ultra/xhigh map to max, medium to high, minimum/light to low, none disables thinking; unknown values return HTTP 400. Thinking-disabled K3 and K2.8 requests are served by K2.8 without thinking. Changing model or effort invalidates context cache. The exact Code model table gives 1,048,576 context and image/video input; membership/quota rules apply. Maximum output, parameters, standalone weights/licence and exact comparable benchmark scores remain unverified.

Model comparison catalog Benchmark library