List your company

Agentic Coding on R&D Coding Benchmark

80Pass RateOpus 4.6
10.3228.4146.564.59Apr 26, 2026

Evaluation results

MethodLinks
Opus 4.6Thinking mode=true2026.0480
Opus 4.5Thinking mode=true2026.0473
Opus 4.52026.0470
DeepSeek-V4-Pro-MaxReasoning effort=Maximum2026.0467
Sonnet 4.52026.0447
Haiku 4.52026.0413