Model Catalog

Every model, X-rayed.

ModelIntegrity is the engineering due-diligence layer for large language models: architectural disclosure, parameter and training provenance, hardware requirements, and integrity scoring for every model — before you bet your stack on it.

23 open-weight LLMs indexed · deep integrity profiling

Architectural Integrity

What the model actually is, not what the release notes claim.

Hardware Truth

Real VRAM, compute, and deployment costs, verified.

Provenance First

Training data lineage and licensing, on the record.

Filters
23 / 23 models

DeepSeek V3 671B MoE

DeepSeek AI

79

integ.

MoEOpenSwiGLU

Params

671B

Context

128k

Layers

61

Attn

MLA

Hidden

7K

KV Heads

128

Instruction Following
87
Hallucination Cal.
73
Profile

Mistral NeMo 12B Instruct

Mistral AI

76

integ.

DenseOpenSwiGLU

Params

12B

Context

128k

Layers

40

Attn

MHA

Hidden

5K

KV Heads

8

Instruction Following
84
Hallucination Cal.
70
Profile

Qwen 2.5 72B Instruct

Alibaba Cloud

81

integ.

DenseOpenSwiGLU

Params

73B

Context

128k

Layers

80

Attn

GQA

Hidden

8K

KV Heads

8

Instruction Following
89
Hallucination Cal.
75
Profile

Qwen 3 32B Instruct

Alibaba Cloud

84

integ.

DenseOpenSwiGLU

Params

33B

Context

128k

Layers

64

Attn

GQA

Hidden

5K

KV Heads

8

Instruction Following
92
Hallucination Cal.
77
Profile

Command R+ 104B Instruct

Cohere

82

integ.

DenseOpenSwiGLU

Params

104B

Context

128k

Layers

64

Attn

GQA

Hidden

8K

KV Heads

8

Instruction Following
88
Hallucination Cal.
76
Profile

DeepSeek R1 671B MoE

DeepSeek AI

82

integ.

MoEOpenSwiGLU

Params

671B

Context

128k

Layers

61

Attn

MLA

Hidden

7K

KV Heads

128

Instruction Following
85
Hallucination Cal.
79
Profile

Mixtral 8×22B 141B MoE Instruct

Mistral AI

74

integ.

MoEOpenSwiGLU

Params

141B

Context

64k

Layers

56

Attn

GQA

Hidden

6K

KV Heads

8

Instruction Following
83
Hallucination Cal.
70
Profile

Llama 3.2 11B Vision Instruct

Meta AI

78

integ.

DenseOpenSwiGLU

Params

11B

Context

128k

Layers

32

Attn

GQA

Hidden

4K

KV Heads

8

Instruction Following
87
Hallucination Cal.
72
Profile

Gemma 3 12B Instruct

Google DeepMind

82

integ.

DenseOpenGELU

Params

12B

Context

128k

Layers

50

Attn

GQA

Hidden

4K

KV Heads

8

Instruction Following
88
Hallucination Cal.
79
Profile

Gemma 3 4B Instruct

Google DeepMind

78

integ.

DenseOpenGELU

Params

4B

Context

128k

Layers

34

Attn

GQA

Hidden

3K

KV Heads

4

Instruction Following
85
Hallucination Cal.
74
Profile

Gemma 2 27B Instruct

Google DeepMind

78

integ.

DenseOpenGELU

Params

27B

Context

8k

Layers

46

Attn

MHA

Hidden

5K

KV Heads

16

Instruction Following
84
Hallucination Cal.
78
Profile

Gemma 3 27B Instruct

Google DeepMind

85

integ.

DenseOpenGELU

Params

27B

Context

128k

Layers

62

Attn

GQA

Hidden

5K

KV Heads

16

Instruction Following
90
Hallucination Cal.
81
Profile

OLMo 2 32B Instruct

Allen Institute for AI

79

integ.

DenseOpenSwiGLU

Params

32B

Context

4k

Layers

64

Attn

GQA

Hidden

5K

KV Heads

8

Instruction Following
82
Hallucination Cal.
75
Profile

Qwen 3 235B-A22B MoE

Alibaba Cloud

83

integ.

MoEOpenSwiGLU

Params

235B

Context

128k

Layers

94

Attn

GQA

Hidden

7K

KV Heads

4

Instruction Following
91
Hallucination Cal.
78
Profile

Phi-3.5 MoE 16x3.8B Instruct

Microsoft

75

integ.

MoEOpenSiLU

Params

42B

Context

128k

Layers

32

Attn

MHA

Hidden

3K

KV Heads

32

Instruction Following
88
Hallucination Cal.
72
Profile

Mistral Large 2 123B Instruct

Mistral AI

77

integ.

DenseOpenSwiGLU

Params

123B

Context

128k

Layers

88

Attn

GQA

Hidden

12K

KV Heads

8

Instruction Following
86
Hallucination Cal.
74
Profile

Mistral 7B v0.3 Instruct

Mistral AI

73

integ.

DenseOpenSwiGLU

Params

7B

Context

32k

Layers

32

Attn

GQA

Hidden

4K

KV Heads

8

Instruction Following
82
Hallucination Cal.
66
Profile

DeepSeek R1 Distill 32B Qwen

DeepSeek AI

81

integ.

DenseOpenSwiGLU

Params

33B

Context

128k

Layers

64

Attn

GQA

Hidden

5K

KV Heads

8

Instruction Following
83
Hallucination Cal.
77
Profile

Phi-4 14B Instruct

Microsoft

79

integ.

DenseOpenSiLU

Params

15B

Context

16k

Layers

40

Attn

MHA

Hidden

5K

KV Heads

10

Instruction Following
90
Hallucination Cal.
80
Profile

Falcon 180B Chat

TII (UAE)

66

integ.

DenseOpenGELU

Params

180B

Context

4k

Layers

80

Attn

MQA

Hidden

15K

KV Heads

8

Instruction Following
74
Hallucination Cal.
60
Profile

Llama 3.1 70B Instruct

Meta AI

79

integ.

DenseOpenSwiGLU

Params

71B

Context

128k

Layers

80

Attn

GQA

Hidden

8K

KV Heads

8

Instruction Following
88
Hallucination Cal.
71
Profile

Llama 3.1 405B Instruct

Meta AI

79

integ.

DenseOpenSwiGLU

Params

405B

Context

128k

Layers

126

Attn

GQA

Hidden

16K

KV Heads

8

Instruction Following
91
Hallucination Cal.
68
Profile

Llama 3.3 70B Instruct

Meta AI

82

integ.

DenseOpenSwiGLU

Params

71B

Context

128k

Layers

80

Attn

GQA

Hidden

8K

KV Heads

8

Instruction Following
92
Hallucination Cal.
74
Profile
Compare
Select model
vs
Select model
Compare