List your company

Web Agent Tasks on Mind2Web Cross-Task

53.2Step Success RateMINDACT
-2.12812.23626.640.964Jan 17, 2024Jun 10, 2024Nov 2, 2024Mar 27, 2025Aug 19, 2025Jan 11, 2026Jun 6, 2026

Evaluation results

MethodLinks
MINDACTBackbone=Flan-T5-large2024.0253.25980.626—
MindActw/o HTML=false2024.015255.175.7——
MindActrequired HTML text inputs=true2024.115255.175.7——
Self-MAPBackbone=Flan-T5-large2024.0251.758.180.526.6—
MINDACT + FixedBackbone=Flan-T5-large2024.0251.35879.726.4—
SynapseBackbone=Flan-T5-large2024.025057.58223.2—
MINDACT + CARBackbone=Flan-T5-large2024.0247.854.579.519.8—
Self-MAPBackbone=Flan-T5-base2024.0247.156.282.524.7—
ReasoningBankBackbone=Gemini-2.5-pro2025.0945.653.662.75.1—
SKILL.nbZero-shot=true2026.06455263.24.7—
ReasoningBankBackbone=Gemini-2.5-flash2025.0944.952.160.44.8—
ReasoningBankZero-shot=true2026.0644.952.160.44.8—
SynapseBackbone=Gemini-2.5-pro2025.0944.750.1613.6—
No MemoryBackbone=Gemini-2.5-pro2025.0944.449.360.23.5—
AWMBackbone=Gemini-2.5-pro2025.0944.448.661.23.7—
AWMonlineZero-shot=true2026.0643.65056.44—
Magma-8BBackbone=LLaMA3 [92], DoM Tree=true2025.0243.454.879.7——
MINDACT + FixedBackbone=Flan-T5-base2024.0242.65180.818.4—
SynapseBackbone=Flan-T5-base2024.0241.949.679.918.4—
MINDACT + CARBackbone=Flan-T5-base2024.0241.447.878.816.1—
SynapseBackbone=Gemini-2.5-flash2025.0941.24759.53.5—
AWMBackbone=Gemini-2.5-flash2025.094146.356.13.5—
No MemoryBackbone=Gemini-2.5-flash2025.0940.34659.13.3—
CodeActZero-shot=true2026.0640.34659.13.3—
SeeAct (GPT-4V)Backbone=GPT-4V [99], Image=true2025.0240.246.473.4——
OmniParser2024.1139.442.487.6——
GPT-4V-OmniParserBackbone=GPT-4V [99], DoM Tree=true, Image=true2025.0239.442.487.6——
Qwen2-UIXBackbone=Qwen2 [124]2025.0238.243.4———
ShowUIFT?=true2024.1137.239.988.6——
ShowUI †FT?=true, Action history only=true2024.1136.939.788——
MINDACTBackbone=Flan-T5-base2024.0236.643.279.114.2—
GPT-4close-source=true2024.1136.341.660.6——
GPT-4w/o HTML=false2024.0136.241.660.6——
GPT-4-MindActBackbone=GPT-4 [98], Image=true2025.0236.241.660.6——
Qwen2-VL-2BFT?=true2024.1133.237.786.4——
CogAgentBackbone=CogVLM [118]2025.0226.930.2———
DAVELanguage Model Backbone=Llama-3.2-3B-Instruct2025.1226.130.8———
SeeClickw/o HTML=true2024.0125.528.387——
SeeClickFT?=true2024.1125.528.387——
SeeClickBackbone=Qwen-VL [3]2025.0225.528.387——
MiniCPM-V-GUIBackbone=MiniCPM-V [128], DoM Tree=true2025.0220.823.886.8——
SeeAct (Gemini-Pro)Backbone=Gemini-Pro [36], Image=true2025.0219.621.567.7——
DolphinLanguage Model Backbone=Llama-3.2-3B-Instruct2025.1219.624———
ShowUI-ZSFT?=false2024.1118.621.485.2——
CogAgentFT?=true2024.1117.622.453——
MindAct (gen)w/o HTML=false2024.0117.520.252——
GPT-3.5-Turbow/o HTML=false2024.0117.420.356.6——
SigLIP 2Language Model Backbone=Llama-3.2-3B-Instruct2025.1217.320.6———
MAELanguage Model Backbone=Llama-3.2-3B-Instruct2025.1216.821———
Fuyu-8B-GUIBackbone=Fuyu-8B [4], DoM Tree=true2025.0215.619.186.1——
Qwen-VLw/o HTML=true2024.0113.315.986.7——
Qwen-VLFT?=true2024.1113.315.986.7——
Qwen-VLBackbone=Qwen-VL [3]2025.0213.315.986.7——
Pix2StructLanguage Model Backbone=Llama-3.2-3B-Instruct2025.1212.315.8———
AIMv2Language Model Backbone=Llama-3.2-3B-Instruct2025.1211.614.7———
DinoV2Language Model Backbone=Llama-3.2-3B-Instruct2025.1211.413.5———
MiniCPM-V+Backbone=MiniCPM-V [128], DoM Tree=true2025.028.51185.6——
Fuyu-8B+Backbone=Fuyu-8B [4], DoM Tree=true2025.026.68.383.9——
DiTLanguage Model Backbone=Llama-3.2-3B-Instruct2025.124.15.8———
Evolving-RLSkill injection=true2026.052.38———41.99
MINDACTBackbone=GPT-3.52024.021.94.327.61—
Evolving-RLSkill injection=false2026.051.59———35.32
GRPOSkill injection=false2026.051.51———28.79
GRPOSkill injection=true2026.050.87———28.98
Base ModelSkill injection=true2026.050.4———14.92
Memento2026.050.4———15.27
Base ModelSkill injection=false2026.050.24———12.57
ReasoningBank2026.050———11.84
DeBERTaBackbone=DeBERTa2024.02—36.8———
GPT-4VBackbone=GPT-4V [99], Image=true2025.02——20.3——