Trust Index

Who documents what, for enterprise use

One score per model, five families, sixteen 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.

Proprietary models (38)

#	Model	Score	Documented	Data	Sovereignty	Transparency	Enterprise	Licence
1	FLUX.1 ProBlack Forest Labs	34	45 %	100	67	13	—	85
2	Claude Opus 4.7Anthropic	31	45 %	100	43	33	—	85
3	Gemini 3.8 FlashGoogle	31	45 %	100	43	33	—	85
4	Gemini 3.7 FlashGoogle	31	45 %	100	43	33	—	85
5	Gemini 3.1 ProGoogle	31	45 %	100	43	33	—	85
6	GPT-5.6 SolOpenAI	31	45 %	100	43	33	—	85
7	Claude Opus 4.8Anthropic	31	45 %	100	43	33	—	85
8	GPT-5.5OpenAI	31	45 %	100	43	33	—	85
9	Gemini 3.6 FlashGoogle	31	45 %	100	43	33	—	85
10	Gemini 3 FlashGoogle	31	45 %	100	43	33	—	85
11	Claude Sonnet 5Anthropic	31	45 %	100	43	33	—	85
12	GPT-5 miniOpenAI	31	45 %	100	43	33	—	85
13	GPT-5.6 LunaOpenAI	31	45 %	100	43	33	—	85
14	GPT-5.6 TerraOpenAI	31	45 %	100	43	33	—	85
15	GPT-6 AstraOpenAI	31	45 %	100	43	33	—	85
16	Claude Fable 5Anthropic	29	45 %	100	43	13	—	85
17	Claude Opus 4.6Anthropic	29	45 %	100	43	13	—	85
18	Claude Opus 5Anthropic	29	45 %	100	43	13	—	85
19	Gemini 3 ProGoogle	29	45 %	100	43	13	—	85
20	Claude Sonnet 4.6Anthropic	29	45 %	100	43	13	—	85
21	GPT-5OpenAI	29	45 %	100	43	13	—	85
22	o3OpenAI	29	45 %	100	43	13	—	85
23	DALL-E 3OpenAI	29	45 %	100	43	13	—	85
24	OpenAI Voice (Realtime)OpenAI	29	45 %	100	43	13	—	85
25	ElevenLabs v3ElevenLabs	27	45 %	100	43	0	—	85
26	Qwen 3 MaxAlibaba	19	40 %	—	0	33	—	100
27	Qwen3.8-MaxAlibaba	17	40 %	—	0	33	—	85
28	Doubao Seed 2.1 ProByteDance	17	35 %	—	0	33	—	85
29	Seedance 2.0ByteDance	17	35 %	—	0	33	—	85
30	Grok 4xAI	16	50 %	0	10	13	—	85
31	Grok 4.5xAI	16	40 %	—	10	13	—	85
32	Muse Spark 1.1Meta	16	35 %	—	15	13	—	85
33	Midjourney v7Midjourney	15	50 %	0	10	0	—	85
34	Runway Gen-4Runway	15	50 %	0	10	0	—	85
35	Sora 2OpenAI	15	50 %	0	10	0	—	85
36	Qwen3.7 MaxAlibaba	14	40 %	—	0	13	—	85
37	Qwen3.7 PlusAlibaba	14	40 %	—	0	13	—	85
38	Yi-Lightning01.AI	14	40 %	—	0	13	—	85

Open-weight models (119)

Self-hosted: data never leaves, most enterprise criteria depend on your deployment.

#	Model	Score	Documented	Data	Sovereignty	Transparency	Enterprise	Licence
1	Mistral Large 3Mistral AI	94	95 %	100	100	85	100	100
2	Mixtral 8x7BMistral AI	93	95 %	100	100	75	100	100
3	Mistral 7BMistral AI	93	95 %	100	100	75	100	100
4	Devstral SmallMistral AI	93	95 %	100	100	75	100	100
5	EuroLLM 9BUTTER Project	93	95 %	100	100	75	100	100
6	FLUX.2 [klein] 4BBlack Forest Labs	93	95 %	100	100	75	100	100
7	Lucie 7BOpenLLM France	93	95 %	100	100	75	100	100
8	Magistral SmallMistral AI	93	95 %	100	100	75	100	100
9	Ministral 3 14BMistral AI	93	95 %	100	100	75	100	100
10	Ministral 3 3BMistral AI	93	95 %	100	100	75	100	100
11	Ministral 3 8BMistral AI	93	95 %	100	100	75	100	100
12	Mistral Nemo 12BMistral AI	93	95 %	100	100	75	100	100
13	Mistral Small 3.2Mistral AI	93	95 %	100	100	75	100	100
14	StarCoder 2BigCode	93	95 %	100	100	75	100	100
15	Voxtral Mini 3BMistral AI	93	95 %	100	100	75	100	100
16	Voxtral Small 24BMistral AI	93	95 %	100	100	75	100	100
17	TRELLIS.2Microsoft	92	95 %	100	77	100	100	100
18	Mistral Small 3Mistral AI	91	95 %	100	100	60	100	100
19	Mixtral 8x22BMistral AI	91	95 %	100	100	60	100	100
20	Kimi K2Moonshot AI	90	95 %	100	67	100	100	100
21	Kimi K2.7 CodeMoonshot AI	90	95 %	100	67	100	100	100
22	Llama 4 MaverickMeta	89	95 %	100	77	100	100	80
23	Nemotron 3 Nano 30B-A3BNVIDIA	89	95 %	100	77	100	100	80
24	Falcon 3 10BTII	89	95 %	100	77	100	100	80
25	Nemotron 3 Super 120B-A12BNVIDIA	89	95 %	100	77	100	100	80
26	Nemotron 3 Ultra 550B-A55BNVIDIA	89	95 %	100	77	100	100	80
27	Gemma 4 31BGoogle	88	95 %	100	77	75	100	100
28	Gemma 4 26B-A4BGoogle	88	95 %	100	77	75	100	100
29	gpt-oss-120bOpenAI	88	95 %	100	77	75	100	100
30	gpt-oss-20bOpenAI	88	95 %	100	77	75	100	100
31	Granite 4.1 8BIBM	88	95 %	100	77	75	100	100
32	Phi-4Microsoft	88	95 %	100	77	75	100	100
33	Phi-4 ReasoningMicrosoft	88	95 %	100	77	75	100	100
34	OLMo 2 32BAllen AI	88	95 %	100	77	75	100	100
35	Falcon 180BTII	88	95 %	100	77	75	100	100
36	Apertus 8BSwiss AI	88	95 %	100	77	75	100	100
37	ChatterboxResemble AI	88	95 %	100	77	75	100	100
38	Dia 1.6BNari Labs	88	95 %	100	77	75	100	100
39	Gemma 4 12BGoogle	88	95 %	100	77	75	100	100
40	Gemma 4 E2BGoogle	88	95 %	100	77	75	100	100
41	Gemma 4 E4BGoogle	88	95 %	100	77	75	100	100
42	Granite 3.3 8BIBM	88	95 %	100	77	75	100	100
43	Granite 4.1 3BIBM	88	95 %	100	77	75	100	100
44	Kokoro-82MHexgrad	88	95 %	100	77	75	100	100
45	LTX-2.3Lightricks	88	95 %	100	77	75	100	100
46	Mochi 1Genmo	88	95 %	100	77	75	100	100
47	OLMo 3 7BAllen AI	88	95 %	100	77	75	100	100
48	Orpheus 3BCanopy Labs	88	95 %	100	77	75	100	100
49	Parakeet TDT 0.6B v3NVIDIA	88	95 %	100	77	75	100	100
50	Phi-4 MiniMicrosoft	88	95 %	100	77	75	100	100
51	SmolLM3 3BHugging Face	88	95 %	100	77	75	100	100
52	TRELLISMicrosoft	88	95 %	100	77	75	100	100
53	TripoSRStability AI	88	95 %	100	77	75	100	100
54	Whisper large-v3OpenAI	88	95 %	100	77	75	100	100
55	Llama 4 BehemothMeta	87	95 %	100	77	85	100	80
56	Llama 4 ScoutMeta	87	95 %	100	77	85	100	80
57	Qwen2.5 72BAlibaba	87	95 %	100	67	100	100	80
58	DeepSeek-Coder-V2 LiteDeepSeek	87	95 %	100	67	100	100	80
59	GLM-5.3Zhipu AI	87	95 %	100	67	100	100	80
60	Kimi K3Moonshot AI	87	95 %	100	67	100	100	80
61	Kimi K2.6Moonshot AI	86	95 %	100	67	75	100	100
62	Qwen3.5 397B-A17BAlibaba	86	95 %	100	67	75	100	100
63	DeepSeek V4-FlashDeepSeek	86	95 %	100	67	75	100	100
64	GLM-4.5Z.ai	86	95 %	100	67	75	100	100
65	DeepSeek-R1DeepSeek	86	95 %	100	67	75	100	100
66	Qwen3-Coder 480BAlibaba	86	95 %	100	67	75	100	100
67	Qwen3 235B-A22BAlibaba	86	95 %	100	67	75	100	100
68	GLM-4.5 AirZ.ai	86	95 %	100	67	75	100	100
69	Qwen3 32BAlibaba	86	95 %	100	67	75	100	100
70	QwQ 32BAlibaba	86	95 %	100	67	75	100	100
71	Qwen2.5-Coder 32BAlibaba	86	95 %	100	67	75	100	100
72	Ministral 8BMistral AI	86	95 %	100	100	100	100	30
73	CogVideoX 5BZhipu AI	86	95 %	100	67	75	100	100
74	DeepSeek-R1 Distill 32BDeepSeek	86	95 %	100	67	75	100	100
75	DeepSeek-R1 Distill 8BDeepSeek	86	95 %	100	67	75	100	100
76	GLM-5.2Zhipu AI	86	95 %	100	67	75	100	100
77	GLM-ImageZhipu AI	86	95 %	100	67	75	100	100
78	HiDream-I1HiDream	86	95 %	100	67	75	100	100
79	InternLM3 8BShanghai AI Lab	86	95 %	100	67	75	100	100
80	Qwen3 0.6BAlibaba	86	95 %	100	67	75	100	100
81	Qwen3 14BAlibaba	86	95 %	100	67	75	100	100
82	Qwen3 4BAlibaba	86	95 %	100	67	75	100	100
83	Qwen3.6 27BAlibaba	86	95 %	100	67	75	100	100
84	Qwen3.6 35B-A3BAlibaba	86	95 %	100	67	75	100	100
85	Qwen3 8BAlibaba	86	95 %	100	67	75	100	100
86	Qwen-Image 2512Alibaba	86	95 %	100	67	75	100	100
87	Wan 2.2 5BAlibaba	86	95 %	100	67	75	100	100
88	Wan 2.2 A14BAlibaba	86	95 %	100	67	75	100	100
89	Z-Image TurboAlibaba	86	95 %	100	67	75	100	100
90	Gemma 3 27BGoogle	85	95 %	100	77	75	100	80
91	Gemma 3 12BGoogle	85	95 %	100	77	75	100	80
92	Gemma 3n E4BGoogle	85	95 %	100	77	75	100	80
93	Llama 3.3 70BMeta	85	95 %	100	77	75	100	80
94	Gemma 3 4BGoogle	85	95 %	100	77	75	100	80
95	Llama 3.1 Nemotron 70BNVIDIA	85	95 %	100	77	75	100	80
96	Llama 3.1 8BMeta	85	95 %	100	77	75	100	80
97	Llama 3.2 3BMeta	85	95 %	100	77	75	100	80
98	Llama 3.2 1BMeta	85	95 %	100	77	75	100	80
99	ArborStability AI	85	95 %	100	77	75	100	80
100	CodeGemma 7BGoogle	85	95 %	100	77	75	100	80
101	Gemma 3 1BGoogle	85	95 %	100	77	75	100	80
102	Hermes 3 8BNous Research	85	95 %	100	77	75	100	80
103	Krea 2Krea AI	85	95 %	100	77	75	100	80
104	Stable Diffusion 3.5Stability AI	85	95 %	100	77	75	100	80
105	Stable Fast 3DStability AI	85	95 %	100	77	75	100	80
106	GLM-5Zhipu AI	84	95 %	100	67	60	100	100
107	DeepSeek V3.2DeepSeek	84	95 %	100	67	60	100	100
108	DeepSeek V3DeepSeek	84	95 %	100	67	60	100	100
109	DeepSeek R2DeepSeek	84	95 %	100	67	60	100	100
110	DeepSeek Coder V3DeepSeek	84	95 %	100	67	60	100	100
111	CodestralMistral AI	84	95 %	100	100	85	100	30
112	FLUX.1 [dev]Black Forest Labs	82	95 %	100	100	75	100	30
113	FLUX.2 [dev]Black Forest Labs	82	95 %	100	100	75	100	30
114	Command R+Cohere	78	95 %	100	77	75	100	30
115	Aya Expanse 32BCohere	78	95 %	100	77	75	100	30
116	Command R7BCohere	78	95 %	100	77	75	100	30
117	Hunyuan3D 2.1Tencent	76	95 %	100	67	75	100	30
118	HunyuanImage 3.0Tencent	76	95 %	100	67	75	100	30
119	HunyuanVideo 1.5Tencent	76	95 %	100	67	75	100	30

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.

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