Speculation: DeepSeek V4.1 Pro could be a 3.1T-param MoE with 2.6TB disk footprint
teortaxesTex · x · 2026-09-10
User teortaxesTex offers a naive extrapolation suggesting V4.1 Pro is a 3.1T-total-parameter MoE with 30/60B active per token, plus an additional 1.1T-param Engram module, for a 2.6TB total disk footprint. Quoted context notes gray-test fv/pro variants output at only 30 tps and may share the same architecture, with v5 possibly reserved for continual learning. The author thinks this scale plus a few months of RL progress could bury Sol/Fable and approach Astra. Unverified speculation.
Related event: Rumor: DeepSeek Trained a 3T-Parameter Model but Held Back Release(3 posts)→
More from Models
- Kimi K3.1 Pro tops MathArena open models, closing frontier gap to about 6 months — teortaxesTex · 2026-10-26
- Ruff creator Charlie Marsh: frontier models find infinite bugs in any software — charliermarsh · 2026-10-26
- Dev testing one skill across Opus 5.5 and Sol 6.1 finds instruction following 'all over the map' — lucasmeijer · 2026-10-26
- supermemory unveils learner-1, claiming continual learning for any model — julianweisser · 2026-10-26
- Anthropic says Opus 5.5 cuts cost per task by ~40% vs Opus 5 — taufiqintech · 2026-10-26
- AI search APIs ranked: top 12 within 7 points, yet prices differ nearly 7x — FinanceYF5 · 2026-10-26