20-Question Affordance Identification on AFFORDANCE20Q 1.0 (test) benchmark leaderboard · SOTA2 Research

20-Question Affordance Identification on AFFORDANCE20Q 1.0 (test)

100Success Rate (SR)OPTIMAL

Evaluation Results

Method	Links
OPTIMAL2026.06		100	2.5
HUMAN2026.06		64.2	10.7
GEMINI-2.5-PRO2026.06		45.9	18.5
GEMINI-2.5-FLASH2026.06		41.4	17.3
DEEPSEEK-V4-FLASH2026.06		41.3	16.9
DEEPSEEK-V4-PRO2026.06		40.5	18.8
GEMINI-2.5-PRO2026.06		40	12.1
DEEPSEEK-V4-PRO2026.06		38.5	11.2
DEEPSEEK-V4-FLASH2026.06		37.9	10.6
GPT-52026.06		35.5	14.5
MINIMAX-M2.52026.06		35.5	9.3
MINIMAX-M2.52026.06		34.9	17.2
GEMMA-3-12B2026.06		33.7	18.6
QWEN3-8B2026.06		33.4	11.7
GEMINI-2.5-FLASH2026.06		33.4	9.9
QWEN3.5-9B2026.06		33.2	14.3
QWEN3-14B2026.06		32.9	10.9
GPT-52026.06		32.4	7
LLAMA-3.1-8B2026.06		31.3	20
GPT-5-MINI2026.06		31.2	10.9
PHI-4-14B2026.06		30.5	11.9
NEMOTRON-9B2026.06		30	17.4
MINISTRAL-8B2026.06		28	12.6
GEMMA-3-12B2026.06		27.4	15.1
QWEN3-14B2026.06		27.3	18.6
PHI-4-14B2026.06		26.4	11.2
QWEN3-8B2026.06		26.3	16.1
QWEN3.5-9B2026.06		25.1	18.6
FIX20Q2026.06		24.8	20
GPT-5-MINI2026.06		22.8	16.9
LLAMA-3.1-8B2026.06		18.4	14.5
MINISTRAL-8B2026.06		17.7	15.1
NEMOTRON-9B2026.06		14.9	8.6