DeepSeek V4.1 Pro rumored to train on 50T+ pretraining tokens, matching Microsoft's scale
teortaxesTex · x · 2026-10-22
TeortaxesTex speculates that DeepSeek's V4.1 Pro will break through 50T pretraining tokens, noting Microsoft already hit 50T with MAU and claiming OpenAI and Ant are above that. He hopes the real figure is far higher—calling it the first 'Whale' model free from the scars of compute poverty. Unconfirmed rumor.
More from Models
- Anthropic most complete, OpenAI most accurate: 14 models compared on medical benchmarks — ArtificialAnlys · 2026-10-22
- BitKindle: AI bootstraps a full C compiler from a 3,584-byte machine-code seed — dotey · 2026-10-22
- A tokenization-shared open model ladder is sorely missing, and all roads lead to Qwen — bclavie · 2026-10-22
- Chonk DOOM: a DOOM-like shooter built for Mistral Large 4 to beat on its own — GuillaumeLample · 2026-10-22
- Research finds multimodal models rarely actually look at images — MoonL88537 · 2026-10-22
- OpenAI Voice Mode Users Hit Long Wait Queues Even on Pro Plan — deviltalk · 2026-10-22