List your company

Web-based tool-use on Mind2Web

39.84Task PerformanceLAR
12.810419.827726.84533.8623May 18, 2026

Evaluation results

MethodLinks
LARBackbone=Qwen3-8B2026.0539.84-2.9
VanillaBackbone=Qwen3-8B2026.0536.73—
ConciseHintBackbone=Qwen3-8B2026.0535.3311.9
COTBackbone=Qwen3-8B2026.0534.15—
TokenSkipBackbone=Qwen3-8B2026.0531.13-16.6
ACONBackbone=Qwen3-8B2026.0530.77-17.7
LARBackbone=Llama-3.1 8B-Instruct2026.0528.3-20.8
VanillaBackbone=Llama-3.1 8B-Instruct2026.0524.4—
ConciseHintBackbone=Llama-3.1 8B-Instruct2026.0517.33-13
ACONBackbone=Llama-3.1 8B-Instruct2026.0515.63-16.7
TokenSkipBackbone=Llama-3.1 8B-Instruct2026.0514.27-1.9
COTBackbone=Llama-3.1 8B-Instruct2026.0513.85—