List your company

Desktop automation on WindowsAgentArena (WAA) v1 (test)

61Overall ScoreVLAA-GUI
30.52838.43946.3554.261Apr 23, 2026

Evaluation results

MethodLinks
VLAA-GUIMax Steps=100, Backbone=Gemini 3 Flash2026.04613573.387.566.752.483.3
VLAA-GUIMax Steps=50, Backbone=Gemini 3 Flash2026.0460.432.673.387.566.752.475
Agent S3Max Steps=100, Backbone=GPT-52026.0456.6——————
VLAA-GUI - Loop BreakerMax Steps=100, Backbone=Gemini 3 Flash, Ablation=Remove Loop Breaker2026.0455.830.266.783.358.347.683.3
Agent S3Max Steps=50, Backbone=GPT-52026.0454.1——————
VLAA-GUI - Search AgentMax Steps=100, Backbone=Gemini 3 Flash, Ablation=Remove Search Agent2026.0453.92563.37566.747.683.3
VLAA-GUI - Loop BreakerMax Steps=50, Backbone=Gemini 3 Flash, Ablation=Remove Loop Breaker2026.0452.627.966.783.358.338.175
VLAA-GUI - Completeness VerifierMax Steps=50, Backbone=Gemini 3 Flash, Ablation=Remove Completeness Verifier2026.0451.311.673.370.854.261.983.3
VLAA-GUI - Completeness VerifierMax Steps=100, Backbone=Gemini 3 Flash, Ablation=Remove Completeness Verifier2026.0451.313.973.370.854.257.183.3
GTA1-32BMax Steps=100, Backbone=o32026.0451.2——————
UI-TARS-2Max Steps=502026.0450.6——————
VLAA-GUI - Search AgentMax Steps=50, Backbone=Gemini 3 Flash, Ablation=Remove Search Agent2026.0449.418.663.37566.738.166.7
Qwen3-VL-32BMax Steps=502026.0431.719.149.754.221.142.225