Model
DeepSeek
DeepSeek's V3 and R1 are large mixture-of-experts models (V3 is ~671B total / ~37B active) whose open weights are released under the OSI-approved MIT license - an unusually permissive posture that allows commercial use and even distillation, and which has produced a large family of R1 distills into Llama and Qwen bases.
Do you really own it?
Partial
none·limited·partial·substantial·full
Analytical input: AOI C · 64.4/100
Floor-weighted, not averaged. The weakest factor caps the level, because ownership is a conjunction. Transparency is weak, which holds the whole verdict down regardless of the rest.
You can own the R1 weights outright under MIT - run, redistribute and distil them self-hosted, and your data stays on your infrastructure - but the family is not uniformly MIT: V3's weights fall under the separate DeepSeek License Agreement with real use restrictions (military, minors, PII, discrimination, legal-rights automation), so check the per-checkpoint licence before assuming a clean grant. Transparency is the deeper cap that holds ownership at partial: China-aligned censorship is baked in and the training is opaque, so you cannot fully know what you are running. Adopt with eyes open for reasoning and distillation where that is acceptable, keep sensitive data off it, and avoid the hosted service - its Privacy Policy stores data on servers in the People's Republic of China and its Terms put disputes under PRC law.
1
Use and modify freelyCan you run, modify and adapt it with no gate and no field-of-use trap?
StrongR1 weights are MIT - own, redistribute and (explicitly) distil them into smaller task-specific models with no field-of-use limit - but the family is not uniformly MIT: V3's WEIGHTS fall under the separate DeepSeek License Agreement v1.0 (V3 code stays MIT) with real use restrictions (no illegal/military use, no harm to minors, no false-info-to-harm, no PII misuse, no defamation/harassment/discrimination, no automated legal-rights decisions) - not a clean grant. Verify the per-checkpoint licence; R1 distills carry their base-model licence (Qwen = Apache-2.0; Llama = Llama 3.x).
How this scores (AOI sub-dimensions)
Openness3/5how much is released - weights, data, code, licence - and how freelyOpen-weights tier: MIT-licensed weights and open documentation via detailed technical reports, but training data is closed and training code only partial, with evaluation partial.
Legal2/5how permissive and clean the licence is for real commercial useThe permissive MIT license is a genuine plus, but it is outweighed for compliance purposes by the systemic-risk exemption being void, no Article 55 documentation, no copyright policy and no training-content summary - a concrete EU compliance gap for a downstream deployer.
2
TransparencyDo you know what it is: weights, training, behaviour, and legible terms?
WeakYou hold the weights but cannot see into them - China-aligned topic censorship is baked into behaviour, the training data is unreleased and code only partial, and misuse is unbenchmarked with no third-party red-team.
How this scores (AOI sub-dimensions)
Provenance3/5how well we can trace and verify what went into the modelVerified deepseek-ai org on Hugging Face, safetensors distribution with checksums, clear canonical source and no malicious-checkpoint incident on the canonical org (checklist ~5/8).
Governance3/5how accountable and well-documented the publisher isActive, named publisher with a verified org and a track record of technical reports and releases, meeting the score-3 anchor.
3
Doesn't fail youIs it reliable and good enough for the job?
ModerateR1's reasoning is a genuinely strong analysis backbone, but the headline score (64.4) is held down by the baked-in censorship that distorts sensitive-topic coverage.
How this scores (AOI sub-dimensions)
Performance4/5how capable it is relative to its classR1 is a strong reasoning model and V3 is competitive among large open-weight models on public leaderboards.
Operational5/5how practical it is to run, serve and maintain in productionFirst-class ecosystem support: broad serving across vLLM, SGLang, llama.cpp and Ollama, an extensive family of community quantizations and distills, and wide third-party hosting availability shortly after release.
Safety3/5whether misuse risks are evaluated and guardrails are providedReleased as instruct/reasoning variants with documented behaviour, meeting the score-3 anchor, but lighter safety tuning than Western frontier labs makes jailbreaks easier and the model exhibits topic censorship - no first-party guard model ships, so deployers must add their own guardrails.
4
Doesn't extract your dataDoes running it keep your knowledge and data yours?
StrongSelf-hosted, the open weights run entirely on your own infrastructure with no clawback - the China-storage and PRC-governing-law clauses attach to DeepSeek's hosted service, not the local weights.
How this scores
Not a scored AOI dimension. For a self-hosted model, data-control is a structural property of running the weights yourself, strong by default unless the model phones home or the licence claws back rights. For a hosted API this factor is the retention + train-on-inputs + residency read, scored from the binding terms.
How the AOI score is computed
The seven dimensions above, each scored 0 to 5, weighted and summed to the 0 to 100 headline. The score is the analytical input behind the ownership verdict, not the verdict itself.
DimensionScoreWeightPoints
Openness3/50.1810.8
Provenance3/50.169.6
Legal2/50.166.4
Safety3/50.169.6
Performance4/50.1411.2
Operational5/50.1212.0
Governance3/50.084.8
HeadlineC · 64.4/100
Grade ceiling: a hard flag (🚩 EU AI Act systemic-risk model with no Article 55 documentation) caps the grade below the raw band. See the classification matrices.
Sources
Every rating traces to a primary document. Read means the text was verified; unverified means it is known to exist but has not yet been read.
DocumentWhat it grounds
Licenceread2026-07-25
DeepSeek-R1, read verbatim: "This code repository and the model weights are licensed under the MIT License.
Licenceread2026-07-25
DeepSeek-V3 licensing splits code from weights (differs from R1): the code repository is MIT ("Copyright (c) 2023 DeepSeek"), but "The use of DeepSeek-V3 Base/Chat models is subject to the Model License" - the separate "DeepSeek License Agreement, Version 1.0", which is NOT MIT.
Third-party analysisread2026-07-25
DeepSeek variants, MIT weights, hosted-service data-privacy scrutiny and bans, topic censorship aligned with Chinese content rules, and China-origin context.
Model cardread2026-07-25
Canonical, verified deepseek-ai organisation on Hugging Face distributing safetensors with checksums.
Third-party analysisread2026-07-25
DeepSeek-R1 is a strong reasoning model and V3 competitive among large open-weight models on public leaderboards.
Privacy Policyread2026-07-25
DeepSeek hosted-service Privacy Policy, read verbatim: "We store the information we collect in secure servers located in the People's Republic of China." Information is retained "as long as necessary to provide our Services".
Terms of serviceread2026-07-25
DeepSeek hosted-service Terms of Use, read verbatim (Section 9.1 Governing Law): the terms are "governed by the laws of the People's Republic of China in the mainland," with negotiation followed by PRC-court litigation and no arbitration.
Technical_reportread2026-07-25
DeepSeek-V3 Technical Report (arXiv 2412.19437): MoE 671B total / 37B active, MLA + DeepSeekMoE, auxiliary-loss-free load balancing, trained on 14.8T tokens using 2.788M H800 GPU-hours.
Technical_reportread2026-07-25
DeepSeek-R1 Technical Report (arXiv 2501.12948), "Incentivizing Reasoning Capability in LLMs via Reinforcement Learning": reasoning incentivized via pure RL with no human-labelled reasoning trajectories, then distilled into smaller models.