GPT-6.5 'Bel' rumored to be OpenAI's return to aggressive pretraining scaling
VraserX · x · 2026-11-01
Blogger VraserX speculates that the rumored GPT-6.5 "Bel" could be the first model where OpenAI went all-in on pretraining scaling again. If size rumors are even remotely accurate, he expects something very interesting — though this remains unconfirmed speculation.
Related event: Multiple leaks point to OpenAI internal model GPT-6.2 codenamed Betelgeuse(10 posts)→
More from Models
- Debate: If local models get "good enough", OpenAI and Anthropic are toast — TheZachMueller · 2026-11-01
- Qwen 4 launch expected next week: code already in Qwen repo, 27B model inbound — kimmonismus · 2026-11-01
- OpenRouter data shows OpenAI and Anthropic model spending surging over 12 weeks — firstadopter · 2026-11-01
- Dev banned by OpenAI says switching to open weights models was the best thing in 2026 — IsForAt · 2026-11-01
- Leaked call shows Anthropic CEO Dario visibly losing it over local inference and open models — beffjezos · 2026-11-01
- Anthropic's Opus 5.5 reversed OpenAI's enterprise spend-share gains, new data shows — firstadopter · 2026-11-01