According to Leo, OpenAI just finished its next >10T pretrain "Bel"
Summary
OpenAI has reportedly completed pretraining its next large-scale AI model with over 10 trillion tokens, referred to as 'Bel'.
Similar Articles
@OpenAI: An internal version of our next major model produced 10 new results on long-standing open problems in mathematics and t…
OpenAI's internal next major model produced 10 new results on long-standing open problems in mathematics and theoretical computer science, using roughly $2,000 worth of tokens at GPT-5.6 Sol API rates.
We have started our most ambitious pre-training run yet (1 minute read)
Google has started its most ambitious pre-training run yet, indicating a major new AI model is in development.
"OpenAI has overcome their pre-training issues, and a much larger model code named “Doug” is actively in the works"
SemiAnalysis reports that Google DeepMind is no longer a frontier lab following a leadership overhaul and key departures, while noting OpenAI has overcome pre-training issues and is working on a larger model code-named 'Doug'.
ByteDance is at an early stage of training a model with as many as 10 trillion parameters
ByteDance is in the early stages of training a large language model with up to 10 trillion parameters, signaling a massive scale-up in AI development.
@VraserX: Doug could be the biggest AI training run the world has ever seen. If OpenAI pairs that scale with SEAL-style continual…
A tweet speculates that OpenAI's Doug training run, combined with SEAL-style continual learning, could push beyond GPT-6 toward AGI.