Alpha
Contact
Assess · Kimi K2

Can you own it?

Ownership levelPartialnone·limited·partial·substantial·fullAnalytical input C · 62.8/100

This page is a projection of the one entry record, the Use & modify and Transparency factors that Assess covers. The full verdict is set by all four factors together, floor-weighted so the weakest caps the whole.

Which domain expands which factor
  • AssessUse & modify + Transparency
  • ImplementData control + Reliability
  • UseReliability
  • SupportTransparency

Intended & out-of-scope use

Kimi is Moonshot AI's model line; the confirmed open release is Kimi K2, a trillion-parameter Mixture-of-Experts model (1T total, 32B active) built with an explicit agentic, tool-use and coding focus. It is intended for frontier-adjacent assistant, agentic and coding workloads where an open-weight model is wanted over a closed API and the serving infrastructure is affordable.

Deploy the Kimi-K2-Instruct (safety-tuned) variant. Kimi-K2-Base is a foundation checkpoint for research and fine-tuning and is out of scope for customer-facing use. The release is China-based, applies China-aligned topic censorship, and has no companion guard model and no EU AI Act documentation package. High-stakes, autonomous or EU-regulated decision-making is therefore out of scope without the external control stack described under Implement and a self-assembled compliance package.

Known limitations, bias & failure modes

Kimi K2 applies content filtering aligned with Chinese content rules on politically sensitive topics, as other China-based open-weight models do. Account for it, especially on non-coding prompts. Instruct alignment withstands casual jailbreaks, but coverage is lighter than at Western frontier labs and there is no companion guard or classifier model. Kimi-K2-Base has no safety tuning and is a research artifact.

There are no small variants. The smallest viable deployment is a multi-node cluster even in block-FP8 (~1 TB of weights). The offsetting strength is capability: on independent agentic and coding benchmarks Kimi K2 is among the strongest open-weight models.

Openness tier & components

Kimi K2 is open_weights, not open-science. The Base and Instruct weights are downloadable under a permissive modified MIT licence with a detailed model card and technical report, so weights, documentation and licence are open. The 15.5T-token training data and the training code are not released, and evaluation is only partially reproducible, so those components are closed or partial. The model can be run, adapted and redistributed but not reproduced.

License terms & permitted use

Kimi K2 ships under a Modified MIT License: standard MIT permissions (use, modify, redistribute, commercialise) plus one attribution clause. A commercial product or service with more than 100 million monthly active users or more than $20 million USD in monthly revenue must prominently display "Kimi K2" on its user interface. For nearly all adopters the licence works as permissive MIT, and only large-scale deployers trigger the clause. The modification means the licence is not OSI-certified, so verify the LICENSE file per checkpoint. The clause is one input to the legal score of 3.

Supply-chain & provenance

Weights are distributed from the verified moonshotai org on Hugging Face as block-FP8 safetensors (no pickle requirement) with per-file checksums. The checkpoint trust checklist scores 5/8. The missing controls include cryptographic weight signing (Sigstore/model-signing) and SLSA build attestation, so provenance scores 3, short of 4 and 5. No incidents or malicious-mirror findings are on record for the canonical org. The weights are large and distributed in block-FP8, so verify per-file checksums and pin the exact revision on download.

EU AI Act posture

Kimi K2 is a GPAI model. On the active-parameter compute estimate (~3×10²⁴ FLOPs: 32B active × 15.5T tokens) it sits under the 10²⁵-FLOPs systemic-risk threshold. Unlike the largest systemic-risk peers, no Article 55 hard flag applies and it plausibly reaches the Article 53 open-source exemption from the Annex XI/XII technical-documentation duties.

The exemption stays partial for two reasons. The modified MIT licence is not an OSI-certified free/open licence. The surviving obligations, a copyright policy and a public training-content summary, are not published and cannot be reconstructed because the corpus is closed. A China-based provider is unlikely to furnish an EU documentation package, so an EU deployer must self-assemble compliance material and takes on these gaps if it places a derivative on the market.

Benchmarks & evaluation

Kimi K2 is among the strongest open-weight models on agentic and coding benchmarks (for example SWE-bench Verified ≈ 65.8, plus strong tool-use and agentic scores). The figures are aggregated from Moonshot's technical report and independent evaluation (ev-perf, ev-tech-report). Coverage is partial because the results are third-party without independent reproduction and the evaluation harness is only partially open.

How this scores

The ownership factors this domain covers, drawn from the one entry record.

1

Use and modify freelyCan you run, modify and adapt it with no gate and no field-of-use trap?

Strong

Near-MIT weights: standard MIT permission to fine-tune, redistribute and commercialise the Base and Instruct checkpoints, with a single attribution trigger. Only a commercial product or service over 100M monthly active users or over $20M USD monthly revenue must prominently display 'Kimi K2' on its UI. The licence lacks OSI certification, so verify the LICENSE per checkpoint, but for nearly every adopter it is a plain permissive grant.

How this scores (AOI sub-dimensions)
Openness3/5how much is released - weights, data, code, licence - and how freelyOpen-weights tier: Base and Instruct weights are downloadable under a permissive (modified) MIT licence with a solid model card and technical report.
Legal3/5how permissive and clean the licence is for real commercial useThe broadly permissive modified MIT licence is better than a restrictive community licence, but it is not OSI-certified and carries a large-scale attribution clause.
2

TransparencyDo you know what it is: weights, training, behaviour, and legible terms?

Weak

The model cannot be inspected: the technical report documents a 1T/32B MoE trained on 15.5T tokens with MuonClip, but the training corpus and training code are closed, and the weights carry China-aligned topic censorship that cannot be inspected.

How this scores (AOI sub-dimensions)
Provenance3/5how well we can trace and verify what went into the modelDistributed from the verified moonshotai org on Hugging Face as block-FP8 safetensors with checksums, with no canonical-org malicious incident on record (checklist 5/8).
Governance3/5how accountable and well-documented the publisher isActive, accountable publisher (Moonshot AI) with a detailed technical report and a verified hub presence.
What this means for adoptionYou substantially use, modify and commercialise the self-hosted Kimi K2 weights under a near-MIT grant. The sole condition is the 100M-MAU / $20M-revenue trigger to display 'Kimi K2' in your UI. The weights run entirely on your own infrastructure, so your data stays yours. Ownership is partial because of transparency: the 15.5T-token corpus and training code are closed and the model carries China-aligned censorship that cannot be inspected. The weights and the hosted service differ: the platform.kimi.ai API trains on your content by default (opt-out only by enterprise agreement, Singapore law), so self-host if data control matters. Confirm the attribution trigger and the per-checkpoint LICENSE before shipping, and budget multi-node infrastructure to run it at all.

Sources

The same evidence records as the entry sheet. Read means the text was verified; unverified means it is known to exist but not yet read.

Model cardread2026-11-19
The Kimi-K2-Instruct model card on the verified moonshotai Hugging Face org describes block-FP8 safetensors, 1T total and 32B active parameters, a 128K context and the Muon optimiser.
Licenceread2026-11-19
The Kimi K2 LICENSE is standard MIT with one added clause, which opens "Our only modification part is that".
Documentationunverified2026-07-25
The moonshotai/Kimi-K2 repository documents the MoE architecture (384 experts, 8+1 selected), the MuonClip optimiser, block-FP8 weights and deployment on vLLM, SGLang, KTransformers and TensorRT-LLM with an OpenAI/Anthropic-compatible API.
Technical_reportread2026-07-25
Moonshot's arXiv technical report "Kimi K2: Open Agentic Intelligence" (2507.20534) describes a 1T-total / 32B-active MoE trained on 15.5T tokens with the MuonClip optimiser (Muon + QK-clip) and reports agentic and coding results (e.g.
Terms of serviceread2026-11-19
The platform.kimi.ai model-use agreement (page dated "Last Updated: July 30th, 2026") says user content is used to improve the services, with opt-out available only via an enterprise or separate written agreement.
Privacy Policyread2026-11-19
The platform.kimi.ai privacy policy says the hosted service trains on user prompts, audio, images, videos and files by default; the controller is MOONSHOT AI PTE.
Documentationread2026-11-19
The hosted Kimi API Platform notice says that from 2026-10-31 K2 series models are no longer available to newly registered users (existing users unaffected) and Moonshot plans to retire all K2 series models by the end of December 2026.
Third-party analysisunverified2026-07-25
On independent evaluation Kimi K2 is among the strongest open-weight models on agentic and coding benchmarks (e.g.
Third-party analysisunverified2026-07-25
Independent analysis notes Kimi K2, like other China-based open-weight models, applies China-aligned content filtering on politically sensitive topics.
Third-party analysisunverified2026-07-25
Kimi K2 Instruct is safety-tuned but with lighter alignment coverage than Western frontier labs and no companion guard model.
Third-party analysisunverified2026-07-25
No public EU AI Act training-content summary, copyright policy, or provider documentation package is published for Kimi K2, and the training corpus is not released.
Third-party analysisunverified2026-07-25
Kimi K2 is served across vLLM, SGLang, KTransformers and TensorRT-LLM, but its 1T-parameter scale requires multi-GPU / multi-node infrastructure even in block-FP8.