1An OpenAI model has disproved a central conjecture in discrete geometry (opens in new tab)(openai.com)1429tedsanders1mo ago1055Save
2Get 2 months of Codex for your enterprise, free (opens in new tab)(openai.com)2tedsanders1mo ago0Save
3Tau-knowledge: benchmarking agents on real-world knowledge (opens in new tab)(sierra.ai)2tedsanders1mo ago0Save
4Mythos for Offensive Security: XBOW's Evaluation (opens in new tab)(xbow.com)2tedsanders1mo ago0Save
5Why SWE-bench Verified no longer measures frontier coding capabilities (opens in new tab)(openai.com)10tedsanders4mo ago0Save
6METR estimates that GPT-5.2 has a 50%-time-horizon of around 6.6 hrs (opens in new tab)(twitter.com)2tedsanders4mo ago0Save
