OpenAI: GPT-5.1-Codex-Max

GPT-5.1-Codex-Max, OpenAI’s latest agentic coding model, demonstrates exceptional reliability with a 99% success rate, consistently providing usable responses. Its speed performance is moderate, ranking in the 27th percentile across benchmarks, while its pricing tends to be premium, positioned in the 6th percentile. The model excels in knowledge-based tasks, achieving perfect accuracy in General Knowledge and Ethics, often being the most accurate at its price point and speed. It also shows strong performance in Instruction Following (95th percentile accuracy) and Coding (93rd percentile accuracy), indicating robust capabilities for software development tasks. Mathematics performance is solid at 93% accuracy. A notable weakness is its performance in Hallucinations, where it achieved 88% accuracy, placing it in the 36th percentile, suggesting room for improvement in acknowledging uncertainty. Email Classification is moderate at 97% accuracy. Overall, GPT-5.1-Codex-Max is a highly reliable model with strong reasoning and coding capabilities, albeit with a premium cost and moderate speed.