List your company

Code Generation on Cangjie-bench

81.94AccuracyILA-agent
0.081621.333342.58563.8367Jan 16, 2026

Evaluation results

MethodLinks
ILA-agentModels=Claude-Sonnet-4.52026.0181.9486.45
ILA-agentModels=Qwen3-Max2026.0173.5580.65
Iterative RAGModels=Claude-Sonnet-4.52026.0165.8170.32
ILA-agentModels=DeepSeek-V3.22026.0163.2370.97
Single-time RAGModels=Claude-Sonnet-4.52026.0159.3563.23
Iterative RAGModels=Qwen3-Max2026.0152.957.42
Iterative RAGModels=DeepSeek-V3.22026.0147.152.23
Single-time RAGModels=Qwen3-Max2026.0146.4550.97
Zero-shotModels=Claude-Sonnet-4.52026.0145.1649.68
Single-time RAGModels=DeepSeek-V3.22026.014046.45
Cangjie GeneratorModels=Specialized2026.0134.8447.1
Zero-shotModels=Qwen3-Max2026.017.749.03
Zero-shotModels=DeepSeek-V3.22026.013.235.16