friday, october 9, 2026
top 10 trending●previous 24h●generated 29d ago
OpenAI launched a decision model called Jev, which TypeSafe claims operates significantly faster and uses fewer tokens than traditional LLMs, reaching a $7.5 billion valuation within weeks of release. The model, based on GPT-6 Luna, is being tested for agent evaluation and judgment tasks, with early comparisons against competitors like Claude, Haiku, and Gemini.
- TypeSafe's Jev valued at $7.5B weeks after launch
- Built on GPT-6 Luna architecture
- Marketed as faster with lower token consumption than LLMs
- Early use cases include agent evaluation and judgment tasks
- Simon Willison released LLM plugin for Jev integration
OpenAI released approximately 700–722 mathematical manuscripts containing novel proofs and counterexamples to long-standing problems, including solutions to the Navier-Stokes equations and the Unique Games Conjecture. Mathematicians expressed awe and anxiety over the scale of the release, with researchers estimating it could take years to fully understand and verify the results.
- OpenAI released 700–722 math manuscripts with novel proofs to previously unsolved problems
- Proofs include solutions to Navier-Stokes equations and the Unique Games Conjecture
- More than three dozen mathematicians described the release as 'unprecedented,' 'overwhelming,' and 'pure insanity'
- Researchers estimate understanding and verifying the full scope could require years of work
- AI demonstrated capability to solve hard problems using methods mathematicians describe as 'inhuman'