moonshotai
MoonshotAI: Kimi K2 0711
moonshotai/kimi-k2
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
Kimi K2, listed on OpenRouter as Kimi K2 0711 and documented upstream as Kimi-K2-Instruct, is a MoonshotAI mixture-of-experts language model with 1 trillion total parameters and 32 billion activated per token. Moonshot describes this post-trained variant as a reflex-grade model for general chat and agentic work, with tool use and autonomous problem-solving as stated design goals. OpenRouter exposes it as text-to-text with a 128K-token context. Vendor-reported evaluations emphasize coding, tool use and math, while OpenRouter's design-arena listings place it mid-table; independent verification is limited.[1][2]
Strengths
- Vendor-reported evaluations emphasize coding, agentic software-engineering and tool-use tasks, matching Moonshot's stated optimization for agentic capabilities; OpenRouter's design-arena results show roughly even win rates across coding design categories.[1][2]
- Large sparse architecture (1T total, 32B activated parameters, 384 experts) trained on 15.5T tokens, which Moonshot says was completed without training instability using its MuonClip optimizer.[2]
- Practical API surface per the OpenRouter catalog: tool calling, common sampling controls, and a 128K-token context window for long inputs.[1][2]
Limitations
- Text-only input and output, and the Instruct variant is described by Moonshot as reflex-grade without long thinking; the OpenRouter listing shows no reasoning mode.[1][2]
- Capability evidence is largely Moonshot's own benchmark reporting plus catalog metadata; independent third-party validation was not included in this source pack, so performance on your workload should be tested directly.[1][2]
- Upstream documentation points to a newer Instruct revision (0905), so this 0711 listing may not reflect the latest release; the catalog states a knowledge cutoff at the end of 2024.[1][2]
Where to use it
- Worth evaluating for agentic coding assistants and tool-using workflows, given the vendor's stated design focus and self-reported results; confirm with your own tasks before adopting.[2]
- A plausible fit for general-purpose chat and instruction-following that benefits from long context and tool-call support, per the catalog and documentation.[1][2]
- Candidate for deployments routed through OpenRouter where structured pricing and endpoint options for this model can be compared in the live tables.[1]
Sources
- [1] OpenRouter catalog · MoonshotAI: Kimi K2 0711Sources checked: Nov 13, 2026
- [2] Model documentation · MoonshotAI: Kimi K2 0711Sources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
Kimi K2 Instruct is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32 billion active per forward pass. It is optimized for...
Specifications
- Model ID
- moonshotai/kimi-k2
- Creator
- moonshotai
- Added to OpenRouter
- 11.07.2025
- Context length
- 131,072
- Modality
- text->text
- Input modalities
- text
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.570 / out $2.3
- Knowledge cutoff
- 31.12.2024
- 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 | — |
| Coding | No measurement | — |
| Agents | No measurement | parameters: tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | — |
| Document analysis | No measurement | — |
| Vision | No measurement | — |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Design Arena | models · codecategorieswin rate 51.7% · rank 124 | 1,024Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · datavizwin rate 49.4% · rank 116 | 1,018Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · gamedevwin rate 46.4% · rank 129 | 964Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 55.1% · rank 118 | 1,021Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · websitewin rate 53.1% · rank 126 | 1,043Elo | CC — below the B band | 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) |
|---|---|---|---|---|---|---|
| Novitafp8 | 131,072 | $0.570 | $2.3 | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.