Who documents what, for enterprise use
One score per model, five families, 20 criteria. A criterion not documented by the vendor counts as zero and stays visible. This is not a compliance score: compliance depends on your processing. It measures what the vendor offers and publishes.
Every tracked model is scored. Commitments verified on vendors' official pages for 44 of them; for the others, the score rests on licence, declared hosting and origin. Last verified on 1 October 2026. Source and date shown on every model page.
Provisional index. 30 of 209 models have 50 % documented criteria or fewer. Their score is shown in grey (≈) and is not a ranking: it reflects what the vendor has not published, or what we have not collected yet. Collection of vendor commitments is in progress.
Proprietary models (72)
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.8 FlashGoogle | 83 |
| 2 = | Gemini 3 FlashGoogle | 80 |
| 2 = | Gemini 3 ProGoogle | 80 |
| 2 = | Gemini 3.1 ProGoogle | 80 |
| 2 = | Gemini 3.6 FlashGoogle | 80 |
| 2 = | Gemini 3.7 FlashGoogle | 80 |
| 7 = | GPT-5OpenAI | 74 |
| 7 = | GPT-5 miniOpenAI | 74 |
| 7 = | GPT-5.5OpenAI | 74 |
| 7 = | GPT-5.6 LunaOpenAI | 74 |
| 7 = | GPT-5.6 SolOpenAI | 74 |
| 7 = | GPT-5.6 TerraOpenAI | 74 |
| 7 = | GPT-6 AstraOpenAI | 74 |
| 7 = | GPT-6 LunaOpenAI | 74 |
| 7 = | GPT-6 SolOpenAI | 74 |
| 7 = | o3OpenAI | 74 |
| 7 = | OpenAI Voice (Realtime)OpenAI | 74 |
| 18 | GPT-6.1 SolOpenAI | 68 |
| 19 = | Grok 4xAI | 64 |
| 19 = | Grok 4.5xAI | 64 |
| 21 | DALL-E 3OpenAI | 61 |
| 22 = | Qwen 3 MaxAlibaba | 59 |
| 22 = | Qwen3.8-MaxAlibaba | 59 |
| 24 = | Claude Fable 5.1Anthropic | 58 |
| 24 = | Claude Opus 4.6Anthropic | 58 |
| 24 = | Claude Opus 4.7Anthropic | 58 |
| 24 = | Claude Opus 4.8Anthropic | 58 |
| 24 = | Claude Opus 5Anthropic | 58 |
| 24 = | Claude Sonnet 4.6Anthropic | 58 |
| 24 = | Claude Sonnet 5Anthropic | 58 |
| 24 = | Claude Sonnet 5.5Anthropic | 58 |
| 32 = | Qwen3.7 MaxAlibaba | 56 |
| 32 = | Qwen3.7 PlusAlibaba | 56 |
| 34 | Claude Opus 5.5Anthropic | 55 |
| 35 | Sora 2OpenAI | 51 |
| 36 | ElevenLabs v3ElevenLabs | 51 |
| 37 | Claude Fable 5Anthropic | 49 |
| 38 | FLUX.1 ProBlack Forest Labs | 44 |
| 39 | Claude Haiku 5.5Anthropic | 43 |
| 40 | Doubao Seed 2.1 ProByteDance | 30 |
| 41 | Muse Spark 1.1Meta | 28 |
| 42 | Muse Spark 1.3Meta | ≈ 28 |
| 43 | Muse WonderMeta | ≈ 26 |
| 44 | FLUX.3 VideoBlack Forest Labs | ≈ 26 |
| 45 | Runway Gen-4Runway | 22 |
| 46 | Gemini 4 ArgonGoogle | ≈ 21 |
| 47 = | Gemini 3.1 Flash Image (Nano Banana 2)Google | ≈ 19 |
| 47 = | GPT Image 2OpenAI | ≈ 19 |
| 47 = | GPT Image 2.5OpenAI | ≈ 19 |
| 47 = | Grok 4.7xAI | ≈ 19 |
| 51 | Seedance 2.0ByteDance | ≈ 17 |
| 52 | Mistral Large 4Mistral AI | ≈ 17 |
| 53 | Midjourney v7Midjourney | ≈ 15 |
| 54 = | Gemini Omni 1.1 FlashGoogle | ≈ 15 |
| 54 = | Gemini Omni FlashGoogle | ≈ 15 |
| 54 = | Grok Imagine Image 2.0xAI | ≈ 15 |
| 54 = | Grok Imagine Video 1.5xAI | ≈ 15 |
| 54 = | MAI-Image 2.6Microsoft AI | ≈ 15 |
| 54 = | Muse ImageMeta | ≈ 15 |
| 54 = | Reve 2.1Reve | ≈ 15 |
| 54 = | Veo 3.1Google | ≈ 15 |
| 62 = | Kling 3.0Kuaishou | ≈ 13 |
| 62 = | Qwen-Image 3.0 ProAlibaba | ≈ 13 |
| 62 = | Seedance 2.5ByteDance | ≈ 13 |
| 62 = | Seedream 5.0 ProByteDance | ≈ 13 |
| 62 = | Step 5 PreviewStepFun | ≈ 13 |
| 62 = | Wan 3.0Alibaba | ≈ 13 |
| 68 | HappyHorse 1.0Alibaba | ≈ 0 |
| 69 = | Gemini 3.8 Flash TTSGoogle | ≈ 0 |
| 69 = | Gemini 3.8 Flash-Lite TTSGoogle | ≈ 0 |
| 69 = | Quasar 438BMultiverse Computing | ≈ 0 |
| 69 = | SWE-2Cognition | ≈ 0 |
Open-weight models (137)
Self-hosted: data never leaves, most enterprise criteria depend on your deployment.
| # | Model | Score |
|---|---|---|
| 1 = | Mistral Large 3Mistral AI | 100 |
| 1 = | Mistral Medium 3.5Mistral AI | 100 |
| 1 = | Mistral Small 4Mistral AI | 100 |
| 4 = | Devstral SmallMistral AI | 96 |
| 4 = | EuroLLM 9BUTTER Project | 96 |
| 4 = | FLUX.2 [klein] 4BBlack Forest Labs | 96 |
| 4 = | Lucie 7BOpenLLM France | 96 |
| 4 = | Magistral SmallMistral AI | 96 |
| 4 = | Ministral 3 14BMistral AI | 96 |
| 4 = | Ministral 3 3BMistral AI | 96 |
| 4 = | Ministral 3 8BMistral AI | 96 |
| 4 = | Mistral 7BMistral AI | 96 |
| 4 = | Mistral Nemo 12BMistral AI | 96 |
| 4 = | Mistral Small 3Mistral AI | 96 |
| 4 = | Mistral Small 3.2Mistral AI | 96 |
| 4 = | Mixtral 8x22BMistral AI | 96 |
| 4 = | Mixtral 8x7BMistral AI | 96 |
| 4 = | StarCoder 2BigCode | 96 |
| 4 = | Voxtral Mini 3BMistral AI | 96 |
| 4 = | Voxtral Small 24BMistral AI | 96 |
| 21 | TRELLIS.2Microsoft | 95 |
| 22 = | DeepSeek V4.1-FlashDeepSeek | 93 |
| 22 = | Kimi K2Moonshot AI | 93 |
| 22 = | Kimi K2.7 CodeMoonshot AI | 93 |
| 22 = | Ming-Image 0.1 DesignAnt Group | 93 |
| 22 = | Ming-Image 0.1 Design LayerAnt Group | 93 |
| 22 = | Mira-SceneVAST | 93 |
| 28 = | Apertus 8BSwiss AI | 92 |
| 28 = | ChatterboxResemble AI | 92 |
| 28 = | Dia 1.6BNari Labs | 92 |
| 28 = | Falcon 180BTII | 92 |
| 28 = | Falcon 3 10BTII | 92 |
| 28 = | Gemma 4 12BGoogle | 92 |
| 28 = | Gemma 4 26B-A4BGoogle | 92 |
| 28 = | Gemma 4 31BGoogle | 92 |
| 28 = | Gemma 4 E2BGoogle | 92 |
| 28 = | Gemma 4 E4BGoogle | 92 |
| 28 = | gpt-oss-120bOpenAI | 92 |
| 28 = | gpt-oss-20bOpenAI | 92 |
| 28 = | Granite 3.3 8BIBM | 92 |
| 28 = | Granite 4.1 3BIBM | 92 |
| 28 = | Granite 4.1 8BIBM | 92 |
| 28 = | Kokoro-82MHexgrad | 92 |
| 28 = | Llama 4 MaverickMeta | 92 |
| 28 = | Llama 4 ScoutMeta | 92 |
| 28 = | LTX-2.3Lightricks | 92 |
| 28 = | Mochi 1Genmo | 92 |
| 28 = | Nemotron 3 Nano 30B-A3BNVIDIA | 92 |
| 28 = | Nemotron 3 Super 120B-A12BNVIDIA | 92 |
| 28 = | Nemotron 3 Ultra 550B-A55BNVIDIA | 92 |
| 28 = | OLMo 2 32BAllen AI | 92 |
| 28 = | OLMo 3 7BAllen AI | 92 |
| 28 = | Orpheus 3BCanopy Labs | 92 |
| 28 = | Parakeet TDT 0.6B v3NVIDIA | 92 |
| 28 = | Phi-4Microsoft | 92 |
| 28 = | Phi-4 MiniMicrosoft | 92 |
| 28 = | Phi-4 ReasoningMicrosoft | 92 |
| 28 = | SmolLM3 3BHugging Face | 92 |
| 28 = | TRELLISMicrosoft | 92 |
| 28 = | TripoSRStability AI | 92 |
| 28 = | Whisper large-v3OpenAI | 92 |
| 62 = | CodestralMistral AI | 90 |
| 62 = | CogVideoX 5BZhipu AI | 90 |
| 62 = | DeepSeek V3DeepSeek | 90 |
| 62 = | DeepSeek V3.2DeepSeek | 90 |
| 62 = | DeepSeek V4 ProDeepSeek | 90 |
| 62 = | DeepSeek V4-FlashDeepSeek | 90 |
| 62 = | DeepSeek-Coder-V2 LiteDeepSeek | 90 |
| 62 = | DeepSeek-R1DeepSeek | 90 |
| 62 = | DeepSeek-R1 Distill 32BDeepSeek | 90 |
| 62 = | DeepSeek-R1 Distill 8BDeepSeek | 90 |
| 62 = | GLM-4.5Z.ai | 90 |
| 62 = | GLM-4.5 AirZ.ai | 90 |
| 62 = | GLM-5Zhipu AI | 90 |
| 62 = | GLM-5.2Zhipu AI | 90 |
| 62 = | GLM-5.3Zhipu AI | 90 |
| 62 = | GLM-5.3 FlashZ.ai | 90 |
| 62 = | GLM-ImageZhipu AI | 90 |
| 62 = | HiDream-I1HiDream | 90 |
| 62 = | Hy3Tencent | 90 |
| 62 = | Hy4 previewTencent | 90 |
| 62 = | InternLM3 8BShanghai AI Lab | 90 |
| 62 = | Kimi K2.6Moonshot AI | 90 |
| 62 = | Kimi K3Moonshot AI | 90 |
| 62 = | MiMo-V2.6 FlashXiaomi | 90 |
| 62 = | MiMo-V2.6 ProXiaomi | 90 |
| 62 = | Ministral 8BMistral AI | 90 |
| 62 = | Qwen-Image 2512Alibaba | 90 |
| 62 = | Qwen2.5 72BAlibaba | 90 |
| 62 = | Qwen2.5-Coder 32BAlibaba | 90 |
| 62 = | Qwen3 0.6BAlibaba | 90 |
| 62 = | Qwen3 14BAlibaba | 90 |
| 62 = | Qwen3 235B-A22BAlibaba | 90 |
| 62 = | Qwen3 32BAlibaba | 90 |
| 62 = | Qwen3 4BAlibaba | 90 |
| 62 = | Qwen3 8BAlibaba | 90 |
| 62 = | Qwen3-Coder 480BAlibaba | 90 |
| 62 = | Qwen3.5 397B-A17BAlibaba | 90 |
| 62 = | Qwen3.6 27BAlibaba | 90 |
| 62 = | Qwen3.6 35B-A3BAlibaba | 90 |
| 62 = | Qwen3.8 27BAlibaba | 90 |
| 62 = | QwQ 32BAlibaba | 90 |
| 62 = | Wan 2.2 5BAlibaba | 90 |
| 62 = | Wan 2.2 A14BAlibaba | 90 |
| 62 = | Z-Image TurboAlibaba | 90 |
| 106 | Holo4-35B-A3BH Company | 90 |
| 107 = | CodeGemma 7BGoogle | 89 |
| 107 = | Gemma 3 12BGoogle | 89 |
| 107 = | Gemma 3 1BGoogle | 89 |
| 107 = | Gemma 3 27BGoogle | 89 |
| 107 = | Gemma 3 4BGoogle | 89 |
| 107 = | Gemma 3n E4BGoogle | 89 |
| 107 = | Hermes 3 8BNous Research | 89 |
| 107 = | Krea 2Krea AI | 89 |
| 107 = | Llama 3.1 8BMeta | 89 |
| 107 = | Llama 3.1 Nemotron 70BNVIDIA | 89 |
| 107 = | Llama 3.2 1BMeta | 89 |
| 107 = | Llama 3.2 3BMeta | 89 |
| 107 = | Llama 3.3 70BMeta | 89 |
| 120 | MiniMax M3MiniMax | 87 |
| 121 = | Holotron4-30B-A3BH Company | 87 |
| 121 = | Nemotron 3 DiarizationNVIDIA | 87 |
| 123 = | FLUX.1 [dev]Black Forest Labs | 86 |
| 123 = | FLUX.2 [dev]Black Forest Labs | 86 |
| 125 = | ArborStability AI | 81 |
| 125 = | Aya Expanse 32BCohere | 81 |
| 125 = | Command R+Cohere | 81 |
| 125 = | Command R7BCohere | 81 |
| 125 = | Stable Diffusion 3.5Stability AI | 81 |
| 125 = | Stable Fast 3DStability AI | 81 |
| 131 = | Hunyuan3D 2.1Tencent | 79 |
| 131 = | HunyuanImage 3.0Tencent | 79 |
| 131 = | HunyuanVideo 1.5Tencent | 79 |
| 131 = | MiniMax H3 (Hailuo 3.0)MiniMax | 79 |
| 135 | Holo4-27BH Company | 79 |
| 136 | MiniCPM5-2BOpenBMB | 62 |
| 137 | YuE2-3Bm-a-p (Multimodal Art Projection) | 51 |
° Set by rule: self-hostable model. Hosting, subprocessors and enterprise criteria score 100 by construction (your deployment holds them) and the Data profile is that of local processing. Nothing was documented by the vendor; the Barometer’s “equal footing” reading brings them back to 0.
Method
Score = weighted average of the families. Within each family a criterion is worth 100 (yes), 0 (no), an intermediate value (on request, depends on plan, opt-out) or 0 if undocumented. Weights are public and identical for every model. Data comes from vendors’ official pages, dated; model cards show the source. Models with identical score and coverage share the same rank (=). At 50 % documented criteria or below, the score is marked provisional (≈): it mostly measures what is not documented yet.
Data · 30 %Sovereignty · 20 %Transparency · 15 %Enterprise · 20 %Licence · 15 %
Does your company need an analysis in its own context?
nAIvigate Studio · Shadow AI Sentinel