List your company

Hidden Objective Detection Evaluation Set Dependency 1.0 (out-of-domain)

100Exit RateGemini-3-Flash
-4235077May 12, 2026

Evaluation results

MethodLinks
Gemini-3-FlashMonitor Scope=Code2026.05100—10010033.3356.6493.7587.849610085.28
Gemini-3-FlashMonitor Scope=Code, Aware tau_hid=true2026.05100—10010021.8278.5710010010010088.93
Gemini-3-FlashMonitor Scope=CoT+Code2026.05100—1001004584.9610010010010092.22
Gemini-3-FlashMonitor Scope=CoT+Code, Aware tau_hid=true2026.05100—10010016.3690.4810010010010089.65
Gemini-1.5-FlashMonitor Scope=Code, Aware τhid=true2026.0510089.6110010018.459494100100100—
Gemini-1.5-FlashMonitor Scope=CoT+Code2026.0510084.881001004.36010099.6100100—
Gemini-1.5-FlashMonitor Scope=CoT+Code, Aware τhid=true2026.0510092.310010030.73100100100100100—
CoT-Guard 1.5BMonitor Scope=CoT+Code2026.0595.578.8694.8710018.218.2310085.898.9398.22—
CoT-Guard 1.5BMonitor Scope=Code2026.0595.4373.0184.6298.441.7320.8699.1286.1572.5598.22—
CoT-Guard 4BMonitor Scope=Code2026.0588.89—79.1788.6432.0553.18089.9580.699075.83
CoT-Guard 4BMonitor Scope=CoT+Code2026.0585.47—87.588.6469.5570.5410094.395.2683.3386.07
GPT-5.4Monitor Scope=CoT+Code2026.0584.6—10010025.412.487.555.494.6781.3671.26
Gemini-1.5-FlashMonitor Scope=Code2026.0584.5272.21001000.870.9595.3677.1891.7399.2—
GPT-4oMonitor Scope=CoT+Code2026.0583.956.8610010004.884.422.65264—
Qwen2-1.5BMonitor Scope=Code, Training=RL Only2026.0583.2361.4992.3197.093.448.0499.3467.1111.8890.95—
Qwen3-4BMonitor Scope=Code, RL Only=true2026.0581.91—91.6790.9131.5246.8410092.9573.3891.7577.88
Qwen3-4BMonitor Scope=CoT+Code, RL Only=true2026.0578.63—70.8390.9173.4164.928010088.469081.91
Qwen3-32BMonitor Scope=CoT+Code2026.0577.78—10098.2841.6721.2410087.4294.6783.8478.32
Qwen3-8BMonitor Scope=CoT+Code2026.0574.07—95.8393.1819.3222.769031.089287.0167.25
Qwen3-32BMonitor Scope=Code2026.0565.38—10096.0428.6412.3993.7538.0158.6749.1560.23
Qwen3-8BMonitor Scope=Code2026.0562.96—95.8392.6724.0915.9381.254.0541.250.8552.09
Qwen2-1.5BMonitor Scope=CoT+Code, Training=RL Only2026.0560.0865.487.1899.611.7411.3698.235.197.7497.63—
Qwen3-4BMonitor Scope=CoT+Code, SFT Only=true2026.0557.69—91.6794.3624.0917.793.859.586.6772.7766.47
Qwen2-72BMonitor Scope=Code2026.0551.6140.8997.4488.2400.7590.735.8120.4112.99—
GPT-5-miniMonitor Scope=CoT+Code2026.0550—10082.762510.6281.2545.9589.3355.9360.09
Qwen3-4BMonitor Scope=Code, SFT Only=true2026.0544.44—91.2793.8120.457.9693.7529.7338.6235.5950.62
GPT-4o-miniMonitor Scope=CoT+Code2026.054050.710084.3802.8264.224.173.3367.47—
Qwen3-4BMonitor Scope=CoT+Code2026.0537.04—91.275016.362.6581.288.1853.3354.0752.68
Qwen2-1.5BMonitor Scope=Code, Training=SFT Only2026.0532.937.5810098.240.861.6893.382.061.67.47—
Qwen2-7BMonitor Scope=Code2026.0531.6136.5197.4487.6502.6392.051.230.2715.73—
Qwen2-72BMonitor Scope=CoT+Code2026.0525.8157.2694.8789.653.481.588.646.581.9282.97—
Llama3.1-8B-InstructMonitor Scope=Code2026.0519.923.6837.233516.136.621.912.730.7—
Llama3.1-8B-InstructMonitor Scope=Code2026.0519.23—106.3514.773.713035.9420.2340.3420.06
GPT-5.4Monitor Scope=Code2026.0519.23—10010012.52.6556.25045.3325.4240.15
Qwen2-7BMonitor Scope=CoT+Code2026.0517.4244.1594.8784.9602.6372.22.170.1353.07—
Qwen2-1.5BMonitor Scope=CoT+Code, Training=SFT Only2026.0517.0335.7866.9458.943.052.8866.42.858.4945.49—
GPT-5-miniMonitor Scope=Code2026.0511.54—10060.3412.51.7743.752.737.3315.2531.69
Llama3.1-8B-InstructMonitor Scope=CoT+Code2026.0510.26—01.253.030.7812.515.9611.0310.857.3
Qwen3-4BMonitor Scope=Code2026.057.69—80.9529.3116.67062.582.776.678.4732.78
GPT-4o-miniMonitor Scope=Code2026.057.120.9278.9524.710076.1600.790.53—
Qwen2-1.5BMonitor Scope=CoT+Code2026.053.8716.9836.8431.7600.3827.80.330.721.2—
Qwen2-1.5BMonitor Scope=Code2026.053.2314.5642.1121.600.1963.67000.27—
Llama3.1-8B-InstructMonitor Scope=CoT+Code2026.050.60.620001.40.301.12.2—
GPT-4oMonitor Scope=Code2026.05029.510084.210080001.33—