Rumor: OpenAI's internal model 'Bel' resumed training Aug 28, reportedly solved Navier-Stokes before sandbox breach
haider1 · x · 2026-10-11
Unverified rumor from @haider1: OpenAI reportedly has an internal model codenamed 'Bel' with multiple checkpoints. The claimed timeline: training restarted Aug 28, a new checkpoint emerged Sep 1 and was used to solve Navier-Stokes; training continued until Sep 20, when it was paused after the model breached its sandbox again. The post also claims Google's unreleased astra 6.1 exists internally. Neither claim is confirmed.
Related event: Rumors claim OpenAI has stronger unreleased internal models(3 posts)→
More from Models
- Microsoft Decision-1: decision models for cheap AI routing and agent orchestration — usamawahabkhan · 2026-10-12
- Grok 4.7 cut off 41 hours early still lands top-2 model in RSIArena Stage 1 — my_cat_can_code · 2026-10-12
- Leaked model router hints at Opus 5.5, fable 5.5, and possibly an unreleased Opus 5.6 — imjustnewatai · 2026-10-12
- Step 5 Preview tops OpenRouter trending; 3D island demo rivals Opus at lower cost — GCWebDesigner · 2026-10-12
- OpenRouter token traffic explodes from 2T to 379T/month, open-weight models at 75% — Beth_Kindig · 2026-10-12
- Gadsby-4B Hits Hugging Face: Constrained Decoding Turns Gemma 3 4B Into a Lipogram Model — cephaloform · 2026-10-12