Autonomous Agent Performance on PRDBench
84.67Success RateOneManCompany (Claude Code Sonnet 4.6 + Gemini 3.1 Flash Lite Preview)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OneManCompany (Claude Code Sonnet 4.6 + Gemini 3.1 Flash Lite Preview)AgentType=Multi-agent2026.04 | 84.67 | 345.59 | |
| Claude-4.5AgentType=Minimal2026.04 | 69.19 | — | |
| GPT-5.2AgentType=Minimal2026.04 | 62.49 | — | |
| CodeXAgentType=Commercial2026.04 | 62.09 | — | |
| Claude CodeAgentType=Commercial2026.04 | 56.65 | — | |
| Qwen3-CoderAgentType=Minimal2026.04 | 43.84 | — | |
| DeepSeek-V3.2AgentType=Minimal2026.04 | 40.11 | — | |
| Qwen CodeAgentType=Commercial2026.04 | 39.91 | — | |
| GLM-4.7AgentType=Minimal2026.04 | 38.39 | — | |
| Gemini-3-ProAgentType=Minimal2026.04 | 22.76 | — | |
| Kimi-K2AgentType=Minimal2026.04 | 20.52 | — | |
| Minimax-M2AgentType=Minimal2026.04 | 17.6 | — | |
| Gemini CLIAgentType=Commercial2026.04 | 11.29 | — |