List your company

Industrial Anomaly Detection on MMAD one-shot 1.0 (test)

95.24Anomaly Discrimination ScoreHuman (expert)
48.148860.374472.684.8256Feb 10, 2026

Evaluation results

MethodLinks
Human (expert)Type=Human2026.0295.247592.3183.3394.286.1180.3786.65
Human (ordinary)Type=Human2026.0286.966.2585.5871.2581.5289.5869.7278.69
Gemini-1.5-flashType=Commercial2026.0283.3869.8863.376.4181.5794.048278.65
Gemini-1.5-proType=Commercial2026.0283.0773.8667.279.9786.2794.8883.0881.19
Reason-IAD (Qwen3-VL)Type=Reason-IAD, Backbone=Qwen3-VL, Scale=8B2026.0275.2475.0762.7876.3885.5896.3184.6279.43
Reason-IAD (Qwen3-VL)Type=Reason-IAD, Backbone=Qwen3-VL, Scale=4B2026.0275.0573.7266.4876.8382.3394.984.4179.1
Qwen3-VLType=Open Source, Scale=4B2026.0274.0259.2562.7470.0979.2292.4783.5974.48
Qwen3-VLType=Open Source, Scale=8B2026.0273.0862.3360.3469.1580.0992.5384.0174.51
GLM-4.1V-ThinkingType=Open Source, Scale=9B2026.0272.3172.0166.4779.9884.6193.7983.2678.92
Qwen2.5-VLType=Open Source, Scale=7B2026.0271.7955.2368.9764.5478.992.5584.6672.38
Reason-IAD (Qwen2.5-VL)Type=Reason-IAD, Backbone=Qwen2.5-VL, Scale=7B2026.0270.7467.9760.8272.7882.6297.1784.9976.73
Qwen3-VLType=Open Source, Scale=2B2026.0270.2950.4249.3564.6978.0991.4580.4369.24
GPT-4oType=Commercial2026.0268.6365.855.6273.2183.4194.9882.874.92
InternVL2Type=Open Source, Scale=76B2026.0268.2554.2256.6666.380.4786.482.9270.75
Kimi-VLType=Open Source, Scale=16B2026.0266.5771.7261.8879.7882.9392.1886.5177.37
AnomalyGPTType=Open Source, Scale=7B2026.0265.5727.4927.9736.8632.1129.8435.8236.52
GPT-4o-miniType=Commercial2026.0264.167.3569.0779.0286.7293.9683.3777.65
Qwen2.5-VLType=Open Source, Scale=3B2026.0263.0347.4754.8766.6680.5786.8582.8468.87
Reason-IAD (Qwen3-VL)Type=Reason-IAD, Backbone=Qwen3-VL, Scale=2B2026.0261.970.254.6371.5179.9294.5981.6173.48
AnomalyR1Type=Open Source, Scale=3B2026.0260.6263.5670.1480.4785.2892.4886.1576.96
Claude-3.5-sonnetType=Commercial2026.0260.1460.1448.8167.1379.1185.1979.8368.36
InternVL2Type=Open Source, Scale=8B2026.0259.9743.8547.9157.678.174.1880.3763.14
LLaVA-NeXTType=Open Source, Scale=34B2026.0257.9248.7952.8771.3480.2881.1277.867.16
LLaVA-NeXTType=Open Source, Scale=7B2026.0257.6433.7947.7251.8467.9381.3974.9159.32
Qwen3-VL-thinkingType=Open Source, Scale=8B2026.0252.3141.540.3745.6736.0264.3453.347.64
LLaVA-OneVisionType=Open Source, Scale=7B2026.0251.7746.1341.8562.1969.7390.3180.9363.27
LLaVA-1.5Type=Open Source, Scale=7B2026.0251.3337.0436.6250.669.7968.2969.5354.74
Random ChanceType=Random2026.025025252525252528.57
LLaVA-1.5Type=Open Source, Scale=13B2026.0249.9638.7846.1758.1773.0973.6270.9858.68