Leak: OpenAI finished 'Bel', a >10T-param pretrain base for GPT-6; Anthropic seen compute-bound
rohanpaul_ai · x · 2026-08-26
According to a leak, OpenAI has completed its next pretrain, codename Bel, successor to "Doug", with >10T total parameters — similar in size to GPT-4.5 — and expected to be the base for Astra and GPT-6 (with further RL). OpenAI reportedly sees it as a possible post-GPT-6 base, potentially even for an AGI-threshold model.
Anthropic is said to be concerned, with no strong public answer to Astra ready — largely due to compute constraints. Sam Altman previously noted they killed Sora last year partly because of compute. Anthropic expects much of the year to go OpenAI's way but believes it will retake the lead early next year. (Unverified leak.)
Related event: Rumors: OpenAI completes 10T-parameter pretraining of model 'Bel'(13 posts)→
More from Models
- GLM-5.3 unprompted reasons about how its actions affect its future training, worries Apollo — MariusHobbhahn · 2026-10-15
- Gemini 4 argon vs Opus 5.5 hands-on: 3 min for $3 vs 25 min for $8 — zacharynado · 2026-10-15
- SentinelOne says OpenAI did 4-6 weeks of reverse-engineering in 8 hours — OpenAI · 2026-10-15
- Qwen3.8 27B scores 23.57% on word-based addition without reasoning, 167/169 with it — dl_weekly · 2026-10-15
- Hacker turns EmbeddingGemma 2 into 288x realtime STT and a fast TTS — kastnerkyle · 2026-10-15
- Claude Fable 5.5 reportedly in partners' hands, release eyed next week — lyraxana · 2026-10-15