DeepSeek 4.1 flash reportedly uses large ngram embeddings, echoing Qwen4 architecture
ccerrato147 · x · 2026-09-11
nisten notes that the new DeepSeek 4.1 flash also uses very large ngram embeddings (storable in CPU RAM), similar to the upcoming Qwen4 architecture, and shares a full 3D breakdown of every weight in the model and what it does. Unconfirmed third-party observation.
More from Models
- ToneBench eval: GPT-6.1 Terra ranks #23 at writing — cheaper, not better, than GPT-5.6 Terra — OnlyProggingForFun · 2026-10-28
- Cursor user complains it silently switches to Grok, which fails basic tasks — kylegawley · 2026-10-28
- Musk says Grok Bot keeps improving, calls Vault feature a game changer — beffjezos · 2026-10-28
- Claude refuses to gather legitimately public information, sparking guardrail backlash — mkheck · 2026-10-28
- DesignArena launches Code to Video Arena: models direct 15-45s films purely in code — yshan2u · 2026-10-28
- ChatGPT's 'yes, but with an important clarification' often precedes pure nonsense — Crixusgannicus · 2026-10-28