DeepSeek ships 552B MoE DeepSeek-V4.1-Flash with 1/4 KV cache, phasing out V4-Pro
dl_weekly · x · 2026-09-19
DeepSeek released DeepSeek-V4.1-Flash, a 552B-parameter MoE with an asymmetric Causal Encoder–Decoder architecture (8B active params for input, 16B for output) and native visual understanding.
- Massive cache savings: KV cache needs only 1/4 the HBM and 1/8 the SSD storage vs. the previous generation, sharply cutting agent cache-hit costs
- Replacing the flagship: third-party tests put it ahead of V4-Pro on performance, cost, speed and runtime; from Sept 14 all deepseek-v4-pro requests route to V4.1-Flash at Flash rates until V4.1-Pro launches
- Lower pricing: off-peak rates remain 50% of peak; new pricing effective Sept 10
Live on the DeepSeek API as deepseek-flash; WorkBuddy (CodeBuddy) and OpenCode already support it, with open-source support teased. A third-party weekly reports a 74.2 DeepSWE v1.1 score vs Opus 5's 74.0.
More from Infra
- Big Tech's Future Data Center Leases Top $1.3T, 70% Added in Last Four Quarters — rohanpaul_ai · 2026-11-02
- Daniel Lemire: AI is exploding demand for boring compute, server CPUs to double — lemire · 2026-11-02
- Physical AI's real story: a 4-layer map of 15 companies behind humanoid robots — pdamodaran · 2026-11-02
- Under 5% of US consumers pay for AI, yet compute demand is already forcing usage cuts — NinaDSchick · 2026-11-02
- Sam Altman: AI query energy cost fell 1,000x-100,000x in 3-4 years — rohanpaul_ai · 2026-11-02
- Marvell and Lumentum to merge, targeting $550B AI connectivity market — BenBajarin · 2026-11-02