Web Browsing on BrowseComp+ (test)
56.4AccuracyACE + Optimizer State
Evaluation Results
| Method | Links | |
|---|---|---|
| ACE + Optimizer StateAgent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 56.4 | |
| ACE + Auxiliary LossesAgent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 54.7 | |
| RCL (all primitives)Agent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 51.3 | |
| ACE + Grouped RolloutsAgent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 50.7 | |
| ACEAgent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 50 | |
| GEPAAgent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 48.7 | |
| ACE + BatchingAgent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 48.3 | |
| ACE + Failure ReplayAgent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 47.3 | |
| ACE + Credit AssignmentAgent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 46.3 | |
| SeedAgent Model=GPT-5.4 Nano, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 40.7 | |
| ACE + Failure ReplayAgent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 40 | |
| ACE + Grouped RolloutsAgent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 38.7 | |
| ACE + Auxiliary LossesAgent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 38 | |
| ACEAgent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 37.3 | |
| ACE + Credit AssignmentAgent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 37.3 | |
| RCL (all primitives)Agent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 37.3 | |
| ACE + Optimizer StateAgent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 36 | |
| GEPAAgent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 35.3 | |
| ACE + BatchingAgent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 31.3 | |
| SeedAgent Model=Gemini 3.1 Flash Lite, Optimizer Model=Claude Opus, Optimization Iterations=302026.04 | 28.9 |