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

Long-horizon web tasks on WebTailBench long-horizon tasks

40.2SR@50

Qwen-3-VL-32B

Updated 22d ago

Evaluation Results

Method	Links
Qwen-3-VL-32B 2026.05		40.2	29.2	47.3	32.9	15,487
Qwen-3-VL-32B 2026.05		39.6	29.6	46.8	33.4	12,214
Qwen3-VL-30B-A3B 2026.05		38	28.8	44.9	33.5	15,456
Qwen3-VL-30B-A3B 2026.05		37.4	29.1	44.3	34	12,173
Qwen-3-VL-32B 2026.05		34.8	27.5	41.9	34.3	4,331
ReVision 2026.05		32.8	23.6	40.2	27.4	9,651
Qwen3-VL-30B-A3B 2026.05		32.5	27.1	39.2	34.8	4,309
Qwen-3-VL-8B 2026.05		30.4	28.4	37.1	34	15,392
Qwen-3-VL-8B 2026.05		29.8	28.7	36.6	34.5	12,089
ReVision 2026.05		29.6	28.1	36.3	30.3	11,907
OpenCUA 2026.05		29.5	28.9	36.2	30.2	12,041
ReVision 2026.05		29.3	27.9	36	30	15,362
OpenCUA 2026.05		29.1	28.6	35.8	29.8	15,329
ReVision 2026.05		28.4	25.4	35.2	29.2	6,731
Qwen-3-VL-8B 2026.05		26	26.2	32.8	35.3	4,266
OpenCUA 2026.05		25.8	29.7	32.4	26.4	4,237
UI-TARS-1.5-7B-DPO 2026.05		18.7	28.5	26.8	33.5	15,214
UI-TARS-1.5-7B-DPO 2026.05		18.1	28.8	22.2	34	11,982
UI-TARS-72B-DPO 2026.05		17.4	28.1	19.5	33.2	15,376
UI-TARS-72B-DPO 2026.05		16.9	28.4	18.2	33.7	12,107
UI-TARS-1.5-7B-DPO 2026.05		16.5	26.6	20.4	35	4,169
Kimi-VL-A3B 2026.05		15.9	29.9	18.1	35	15,438
UI-TARS-72B-DPO 2026.05		15.3	26.8	17.4	34.6	4,248
Kimi-VL-A3B 2026.05		14.8	30.3	16.9	35.4	12,217
Kimi-VL-A3B 2026.05		13.2	31	15	36.1	4,318
Qwen2.5-VL-72B 2026.05		9	30.6	10.1	35.7	15,561
Qwen2.5-VL-7B 2026.05		8.6	30.7	10.1	36	15,361
Qwen2.5-VL-72B 2026.05		8.2	31	9.4	36.2	12,244
Qwen2.5-VL-7B 2026.05		7.9	31.1	9.2	36.5	12,067
Qwen2.5-VL-32B 2026.05		7.5	31	8.6	36.1	15,472
Qwen2.5-VL-72B 2026.05		7.5	31.7	8.6	36.9	4,349
Qwen2.5-VL-32B 2026.05		6.8	31.4	7.9	36.8	12,183
Qwen2.5-VL-7B 2026.05		6.5	31.8	7.8	37.2	4,213
Qwen2.5-VL-32B 2026.05		6.2	32.1	7.1	37.4	4,287