Black Forest Labs' image generation and editing model with bounding-box layouts, multiple reference images and 4K output
Kimi K2.8 Preview
New· 4Moonshot AI's coding and agent model for Kimi Code, with a context window up to 1M tokens
About Kimi K2.8 Preview
Members of Kimi Code received a new coding model on 11 September 2026, without changing anything on their side. Moonshot AI rolled out Kimi K2.8 Preview by replacing the model behind the existing model id, kimi-for-coding, so current clients and third-party tools picked it up with no configuration change.
What Moonshot says it improved
Moonshot describes the coding and agent abilities of K2.8 Preview as improved across the board. It says the model thinks significantly more efficiently than its predecessor, K2.7 Code, and that its overall performance sits near that of the flagship Kimi K3. These are Moonshot's own statements: no benchmark figures were published with the release, so there is nothing to check the comparison against.
Thinking effort and input types
The model supports three thinking-effort levels: low, high and max. A user can therefore trade speed and cost against depth of reasoning for a given task. Besides text, it accepts image and video input, which matters for coding work that involves screenshots, interface mockups or recordings of a bug. The context window reaches up to 1M tokens, exactly 1,048,576, large enough to hold substantial codebases and long agent sessions.
Where it is offered
Moonshot's documentation presents K2.8 Preview as a Kimi Code model for members rather than as a separately priced API product. No licence was found on a Moonshot page. The Preview label suggests the model is still being refined, so details may change as Moonshot publishes more about it.
Fit
It suits developers already working in Kimi Code who want a model tuned for coding and agent tasks, and who value a very long context window and image and video input.
TrustList summary
Kimi K2.8 Preview is a model rolled out by Moonshot AI inside Kimi Code on 11 September 2026. It took over the existing model id used by the coding product, so current clients and third-party tools pick it up with no configuration change. It is meant for coding and agent work.
Features
- Coding and agent skills: Moonshot describes these as improved across the board.
- More efficient thinking: Moonshot says it reasons more economically than the earlier K2.7 Code.
- Three effort levels: low, high and max thinking settings.
- Multimodal input: accepts images and video as well as text.
- Large context: up to 1M tokens, or 1,048,576.
Clients
Members of Kimi Code. Moonshot's documentation presents it as a Kimi Code model rather than a separate API product.
Key facts
- Released: 11 September 2026
- Maker: Moonshot AI
- Positioning: Moonshot says overall performance is close to its flagship Kimi K3
- Benchmarks: none were published with the release
Summary written by TrustList from the business's own published information, Nov 2026.
Benchmarks & AI stats
| Benchmark | Official | Community avg |
|---|---|---|
| No benchmark scores yet. Be the first to add one. | ||
“Official” values are editor-approved and feed the ranking. “Community avg” is the mean of member submissions (shown for transparency; it never affects the ranking until an editor approves a value).
Sign in to add a benchmark score for this model.
Kimi K2.8 Preview launches
Trust Score
An earned signal from verification, reviews, awards, transparency and engagement - the vendor can't buy it.
Updated 9/16/2026
What the business tells TrustList
- Founded
- 2026 (0 years) evidence sent
Request a demo or quote from Kimi K2.8 Preview
Reviews
Write the first review of Kimi K2.8 Preview
Used it? Your experience helps other buyers decide.
Write a reviewQuestions & answers
No questions yet. Be the first to ask about Kimi K2.8 Preview.
Kimi K2.8 Preview alternatives
Similar listings buyers compare against this one.
Cantina Security's first open-weight security model, a post-trained GLM-5.3-Flash for vulnerability research and exploit checks
Open-weight translation model from Bilibili's Index team, covering 150 languages and following glossary and style instructions
Open-source 2B decision model from AWS's Strands Agents project that picks options and returns scores with a reliability figure
Anthropic's smallest and fastest Claude 5.5 model, available on its own API and three cloud platforms
Microsoft AI's real-time speech-to-text model for voice agents, transcribing 60 languages with continuous language detection
Aleph Alpha's open-weight English-German mixture-of-experts model for self-hosting, released under Apache 2.0 in October 2026
Perplexity's open-weight decision model that picks among given options for routing and checks in AI systems
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- Write a reviewHelp other buyers decide
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- ArticlesGuides, comparisons and research from the editorial desk and the community.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.