Automating Agent Evaluation on AgentEvalBench
65Eval@1EvalAgent
Evaluation Results
| Method | Links | |
|---|---|---|
| EvalAgentLLM=Sonnet 4.52026.05 | 65 | |
| EvalAgentLLM=Haiku 4.52026.05 | 62.5 | |
| Agent-Sourcecode (B2)LLM=Sonnet 4.52026.05 | 60 | |
| Agent-Sourcecode (B2)LLM=Haiku 4.52026.05 | 45 | |
| Agent-Onestage (B3)LLM=Sonnet 4.52026.05 | 35 | |
| Agent-Twostage (B4)LLM=Haiku 4.52026.05 | 32.5 | |
| Agent-Twostage (B4)LLM=Sonnet 4.52026.05 | 30 | |
| Agent-Onestage (B3)LLM=Haiku 4.52026.05 | 17.5 | |
| LLM-Singleturn (B1)LLM=Sonnet 4.52026.05 | 17.5 | |
| LLM-Singleturn (B1)LLM=Haiku 4.52026.05 | 15 |