Share your thoughts, 1 month free Claude Pro on usSee more

Progress Reasoning on Progress-Bench Cross-View 1.0 (test)

15.2NSE

ProgressLM-3B-RL

Updated 1mo ago

Evaluation Results

Method	Links
ProgressLM-3B-RL 2026.01		15.2	88.8	11.7
ProgressLM-3B-SFT 2026.01		20.4	67.8	8.8
GPT-5 2026.01		20.5	89.4	1.8
GPT-5-mini 2026.01		21.1	89.1	0.4
Qwen3-VL-32B 2026.01		21.9	77.8	0.4
Qwen3-VL-8B 2026.01		25.2	61.3	0
Qwen2.5-VL-72B 2026.01		30.9	50.7	13.4
Qwen2.5-VL-3B 2026.01		33.4	28.9	6.5
Qwen3-VL-4B 2026.01		34.1	51.6	0
Qwen2.5-VL-7B 2026.01		36.5	26.7	26
Intern3.5-VL-38B 2026.01		38.1	49.7	51.3
Intern3.5-VL-4B 2026.01		43.5	12.1	0.2
Qwen2.5-VL-32B 2026.01		45.7	29.4	0
Intern3.5-VL-8B 2026.01		64.1	15.7	0.4
Qwen3-VL-2B 2026.01		66.2	30.9	0
Intern3.5-VL-14B 2026.01		66.4	-40.3	0