List your company

Embodied Spatial Reasoning on S3-Eval (real part)

82.1Overall ScoreHuman level
23.8638.9854.169.22Mar 25, 2026

Evaluation results

MethodLinks
Human level2026.0382.169.190.869.174.791.568.577.591.991.395.675.869.174.391.996.1
AMF-VLM2026.0357.8413748.250.171.845.177.575.559.744.549.770.96464.472
cambrian-strained on our data=true, streaming inference modification=true2026.035138.43341.944.765.138.855.270.954.848.245.458.259.354.856.2
streaming-vlmtrained on our data=true2026.0349.135.230.438.641.362.335.551.967.451.946.142.954.461.852.664.8
cambrian-s2026.0345.139.834.725.34565.940.432.369.946.744.434.960.843.237.264.8
vlm-3r2026.0342.134.128.53033.95330.43959.248.144.44142.852.545.851
GPT-4o2026.0341.728.129.831.525.859.325.539.561.851.444.748.227.854.752.249.5
InternVL2-26B2026.0341.335.829.130.632.556.832.719.458.649.253.242.922.85550.650.1
InternVL2-8B2026.0340.519.727.837.53553.826.638.65847.644.343.924.655.250.442.5
Gemini-1.5-pro2026.0340.420.822.722.939.945.236.729.869.546.744.149.55341.559.137.4
Qwen2-VL-7B2026.033925.327.727.623.757.322.136.159.849.341.648.52551.95047.3
Gemini-1.5-flash2026.0338.224.627.226.722.956.321.936.257.248.241.647.223.151.448.946.6
Qwen2-VL-72B2026.0337.334.125.238.128.25335.629.657.542.954.146.221.948.328.56.3
InternVL2-1B2026.0336.220.536.617.821.851.213.639.162.345.544.233.720.952.24736.1
llava-video2026.0334.914.833.111.521.252.318.632.663.446.94418.217.951.349.338.4
llava-ov-1.5-72B2026.0334.217.327.121.131.149.624.338.453.744.341.219.618.336.644.632.8
llava-ov-1.5-7B2026.0330.11.729.123.811.8440.661.445.539.644.522.13.842.141.224.8
Qwen2-VL-2B2026.0329.923.921.92928.239.518.130.731.725.328.437.72549.526.646
Frequent2026.032925.325.832.63425.933.524.125.926.133.71129.537.526.926.8
Random2026.0326.1—23.1——25.2——25.124.634.4———24.523.5