List your company

Web Navigation on VisualWebArena

54Published VWA Success RateSGV
15.20825.27935.3545.421May 24, 2026May 26, 2026May 29, 2026Jun 1, 2026Jun 3, 2026Jun 6, 2026Jun 9, 2026

Evaluation results

MethodLinks
SGVWhere compute is spent (Eq. 3 term)=Cverify ↑ (two-pass self-grounded verifier), Paid=per-task2026.05542.2
WALTWhere compute is spent (Eq. 3 term)=Cpre ↑ (offline tool discovery, 100 steps/tool), Paid=pre-eval2026.0552.9—
WALTModel=GPT-5, Access=Proprietary2026.0652.9—
WebChallengerModel=GLM-4-32B + Qwen3-VL-4B-Instruct, Access=Open-Source, Evaluation Strategy=zero-shot2026.0648.7—
Mobile-Agent-v3.5Model=Qwen3-VL-32B, Access=Open-Source, Evaluation Strategy=fine-tuned2026.0646.6—
Agent-as-AnnotatorsModel=Qwen3.5-9B, Access=Open-Source, Evaluation Strategy=fine-tuned2026.0633.9—
ExACTWhere compute is spent (Eq. 3 term)=Nrollout ↑ (reflective MCTS), Paid=per-task2026.0533.7—
GenericAgentModel=GPT-4o, Access=Proprietary2026.0626.7—
Tree SearchWhere compute is spent (Eq. 3 term)=Nrollout ↑ (best-first search), Paid=per-task2026.0526.4—
WebDreamerModel=Qwen2-VL-7B, Access=Open-Source, Evaluation Strategy=fine-tuned2026.0621.9—
GenericAgentModel=Claude 3.5 Sonnet, Access=Proprietary2026.0621—
GenericAgentModel=GPT-4o-mini, Access=Proprietary2026.0616.9—
Tree SearchModel=Llama-3-70B-Instruct, Access=Open-Source, Evaluation Strategy=zero-shot2026.0616.7—