Rumors that OpenAI recently finished new >10T parameter training run

OpenAI has completed a new pretraining run codenamed 'Bel', a giant model with over 10 trillion total parameters, similar in size to GPT-4.5, which will serve as the base for Astra and GPT-6 after further reinforcement learning, and potentially for an AGI-threshold model. The news has concerned Anthropic, which believes it has no good response to Astra prepared for public launch due to compute constraints, though it expects to regain its position early next year.

🚨 SCOOP: OpenAI recently finished its next pretrain, codename "Bel" - the successor to "Doug", which is expected to be the base for Astra and GPT-6 w/ further RL Bel is a giant pretrain with 10T total parameters, similar in size to GPT-4.5. OpenAI expect it to be a post-GPT-6 base, and potentially even the base for an AGI-threshold model. All of this has Anthropic concerned. OpenAI believe Anthropic have no good response to Astra prepared for a public launch, and my understanding is that they're correct - mostly as a result of compute constraints, Anthropic are bracing for much of the rest of the year to go OpenAI's way, but think they'll be back on top early next year.