Causal Dynamics Lab outperforms Anthropic and OpenAI in multiple coding tests

Methodology: Benchmarks were run against Claude Code (Opus-4.6) and OpenAI Codex (GPT-5.4) using the publicly available ...