moonshotai
MoonshotAI: Kimi K2.7 Code
moonshotai/kimi-k2.7-code
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
MoonshotAI’s Kimi K2.7 Code is a coding-focused agentic variant built on Kimi K2.6. Moonshot says it targets reliable end-to-end work on long-horizon software tasks and uses about 30% fewer thinking tokens than K2.6. The model uses a large mixture-of-experts design, accepts text and images, returns text, and supports a roughly 256K-token context. Vendor-reported results improve over K2.6 across the listed coding and agentic tests, but each test still trails at least one named comparison system. Evidence comes from OpenRouter and Moonshot documentation; independent verification is limited.[1][2]
Strengths
- Designed for long-horizon, end-to-end software engineering as an agentic coding model built on Kimi K2.6.[2]
- Combines a roughly 256K-token context with text and image inputs, text output, tool calling, structured outputs, and sampling controls.[1][2]
- Moonshot reports roughly 30% lower thinking-token use than K2.6 and higher results than K2.6 on every listed coding and agentic benchmark.[2]
Limitations
- Thinking and preserve_thinking are mandatory, and instant mode is unsupported, so users cannot disable these behaviors.[1][2]
- Video input is described as experimental and limited to Moonshot’s official API, while the OpenRouter listing only specifies text and image inputs.[1][2]
- The source pack contains catalog data plus vendor documentation, but no independent evaluation or hardware requirements; validate accuracy, latency, safety behavior, and cost on target workloads.[1][2]
Where to use it
- Consider for long-context implementation, refactoring, debugging, and maintenance tasks where an agent must sustain multi-step work.[1][2]
- Evaluate for coding assistants that benefit from preserved multi-turn reasoning, tool calls, and structured outputs.[1][2]
- Test on image-informed programming work, such as interpreting screenshots or diagrams; treat video input as an official-API experiment.[1][2]
Sources
- [1] OpenRouter catalog · MoonshotAI: Kimi K2.7 CodeSources checked: Nov 13, 2026
- [2] Model documentation · MoonshotAI: Kimi K2.7 CodeSources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
MoonshotAI: Kimi K2.7 Code is a coding-focused model in Moonshot AI's Kimi K2 family, built to complete end-to-end programming tasks reliably over long contexts. It uses a native multimodal mixture-of-experts...
Specifications
- Model ID
- moonshotai/kimi-k2.7-code
- Canonical ID
- moonshotai/kimi-k2.7-code-20260612
- Creator
- moonshotai
- Added to OpenRouter
- 12.06.2026
- Context length
- 262,144
- Modality
- text+image->text
- Input modalities
- text, image
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.600 / out $3.5
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | No measurement | reasoning: mandatory · parameters: include_reasoning, reasoning |
| Coding | BB — next 30% of the cohort | — |
| Agents | BB — next 30% of the cohort | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: mandatory · parameters: include_reasoning, reasoning |
| Document analysis | No measurement | — |
| Vision | No measurement | input: image |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Artificial Analysis | Coding index | 60.8 | BB — next 30% of the cohort | 13.11.2026 |
| Artificial Analysis | Agentic index | 21 | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | agents · agenticgamedevwin rate 42.8% · rank 22 | 1,079Elo | CC — below the B band | 13.11.2026 |
| Design Arena | agents · androidnativewin rate 46% · rank 31 | 1,129Elo | CC — below the B band | 13.11.2026 |
| Design Arena | agents · fullstackwin rate 50.9% · rank 24 | 1,154Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | agents · godotgamedevwin rate 49.2% · rank 25 | 1,130Elo | CC — below the B band | 13.11.2026 |
| Design Arena | agents · htmlslideswin rate 53.6% · rank 11 | 1,097Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | agents · mobileappswin rate 46.2% · rank 35 | 1,136Elo | CC — below the B band | 13.11.2026 |
| Design Arena | agents · python-pptxslideswin rate 44.6% · rank 24 | 1,114Elo | CC — below the B band | 13.11.2026 |
| Design Arena | agents · webappswin rate 46.9% · rank 31 | 1,158Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · 3dwin rate 49.7% · rank 43 | 1,223Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · asciiartwin rate 47.7% · rank 32 | 1,185Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · codecategorieswin rate 49.7% · rank 45 | 1,244Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · datavizwin rate 48.3% · rank 50 | 1,222Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · gamedevwin rate 46.3% · rank 49 | 1,210Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · svgwin rate 43.8% · rank 44 | 1,166Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 50.8% · rank 42 | 1,252Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · websitewin rate 51.5% · rank 37 | 1,259Elo | AA — next 20% of the cohort | 13.11.2026 |
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| CoreWeaveint4 | 262,144 | $0.600 | $3 | — | — | 100% |
| Inceptronint4 | 262,144 | $0.600 | $3.5 | — | — | 100% |
| StreamLake | 256,000 | $0.712 | $3 | — | — | 100% |
| Veniceint4 | 256,000 | $0.750 | $3.5 | — | — | 100% |
| SiliconFlowfp8 | 262,144 | $0.859 | $3.8 | — | — | 100% |
| Novitaint4 | 262,144 | $0.912 | $3.84 | — | — | 100% |
| GMICloudfp8 | 262,144 | $0.950 | $4 | — | — | — |
| Alibabafp8 | 262,144 | $0.950 | $4 | — | — | 100% |
| Cloudflare | 262,144 | $0.950 | $4 | — | — | 100% |
| Moonshot AI | 262,144 | $0.950 | $4 | — | — | 100% |
| Moonshot AI | 262,144 | $1.9 | $8 | — | — | — |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.