Report: OpenAI finished 'Bel', a >10T-parameter pretrain that could underpin post-GPT-6 systems
Dr_Singularity · x · 2026-08-26
A report claims OpenAI has finished a pretrain codenamed Bel, exceeding 10 trillion parameters, expected to succeed "Doug" and potentially serving as the foundation for post-GPT-6 systems. Unverified — OpenAI has not confirmed.
More from Models
- Together Ranks Top Open Models: Kimi K3 and DeepSeek V4 Lead Use Cases — togethercompute · 2026-08-26
- Questions over Astra's progress: 2 months for 3 more models? — teortaxesTex · 2026-08-26
- View: Tokens-per-second matters more than model size now — natesiggard · 2026-08-26
- Tiel-Coder-35B achieves 121.4 tok/s for local inference — DerTomsn · 2026-08-26
- Benchmark: Tool Calling Performance of Qwen 35B-A3B Variants — OsmanthusBloom · 2026-08-26
- Grok 4.6 now available on OpenCode Go subscription — veggie_eric · 2026-08-26