o3 16 series

Each model's score on the Intelligence board (Text arena) on the last day it was listed

1. o3-2025-04-16No longer on the boardsReleased Apr 23, 20251,432#112

• Feb 232026ResearchOpenAI ditches SWE-bench Verified over benchmark contamination
• Aug 62025ResearchMicrosoft unveils CLIO to give scientists control over AI reasoning
• Jul 242025AI modelsOpenAI reportedly sets GPT-5 launch for early August
• Jun 302025ResearchMicrosoft's AI orchestrator crushed human doctors on complex diagnoses
• Jun 182025AI modelsOpenAI adds guardrails for future bio risks
• Jun 132025ResearchNew paper challenges Apple's claim that AI reasoning collapses