Our new X account is live! Follow @wizwand_team for updates

Physical Reasoning on GameBench 1.0 (test)

56.1AVG Score

GPT-4o

Updated 4d ago

Evaluation Results

Method	Links
GPT-4o 2026.01		56.1	55.4	48.9	60.3	67.3	55.3
Gemini-1.5-pro 2026.01		55.2	55	49.6	58.7	63.3	51.1
Claude3.5-Sonnet 2026.01		54.3	53.7	49.6	54.5	71.4	55.3
Qwen-VL-Max 2026.01		50.9	46.8	52.6	55	69.4	51.1
Gemini-1.5-pro-flash 2026.01		48.5	50.4	41.4	50.3	49	42.6
Qwen2.5-VL-7B (w/ PhysGame) 2026.01		48.1	51.5	42.1	48.1	34.7	44.7
InternVL2.5-8B (w/ PhysGame) 2026.01		48	44.4	43.6	56.6	67.3	40.4
LLaVA-OneVision 2026.01		47.7	46.3	48.1	49.2	59.2	42.6
Claude3.5-SonnetV2 2026.01		47.6	47	42.1	51.9	53.1	46.8
GPT-4V 2026.01		45.9	45.9	43.6	43.9	53.1	53.2
Qwen2.5-VL-7B 2026.01		44.4	46.5	40.6	46	34.7	38.3
Qwen2-VL-7B (w/ PhysGame) 2026.01		43.8	45.5	47.4	48.7	40.8	36.2
GPT-4o-mini 2026.01		40.3	40.7	39.1	41.3	36.7	40.4
InternVL2.5-8B 2026.01		38.6	35.9	33.1	43.4	59.2	40.4
PPLLaVA 2026.01		38.4	37.7	38.3	36	51	42.6
Qwen2-VL-7B 2026.01		37.5	35.5	34.6	37	34.7	34
VideoChat2 2026.01		34.3	33.1	31.6	37	44.9	31.9
InternVL2 2026.01		33.4	34.2	37.6	29.1	30.6	34
ST-LLM 2026.01		32.8	30.3	30.8	38.6	49	23.4
LLaVA-Next-Video 2026.01		32.2	31.8	25.6	37	42.9	23.4
Chat-UniVi 2026.01		29.5	30.7	30.1	30.7	20.4	21.3
Video-LLaVA 2026.01		29	29.2	28.6	31.2	30.6	17