Leak: DeepSeek V4.1 Pro ~2T Params, Skeptical of Kimi's Path
Leaker teortaxesTex claims DeepSeek views Kimi's 2.8T approach as hard to scale and is seeking a cheaper recipe; DeepSeek V4.1 Pro is reportedly 2T params, sparse-trained from scratch without instability.
2026-10-03 ~ 2026-10-03 · 2 related posts
- Report: DeepSeek views Kimi's approach as hard to scale, seeks a cheap scalable recipe — teortaxesTex · 2026-10-03
- Leak: DeepSeek V4.1 Pro ~2T params, trained sparse from scratch with no instabilities — teortaxesTex · 2026-10-03