Releases·Model Releasefeatured
DeepSeek launches V4.1 Flash: 552B MoE with asymmetric architecture, open weights, V4 Pro phase-out
DeepSeek officially released DeepSeek-V4.1-Flash on September 10, 2026, after an internal API beta opened September 8. The company says the 552B-parameter MoE model uses a new Causal-Encoder-Decoder architecture with asymmetric compute (8B activated on input, 16B on output) and comprehensively outperforms V4 Pro on performance, cost, speed, and time-to-completion, prompting an orderly V4 Pro phase-out. Weights and a tech report are on Hugging Face, and Tencent and OpenCode have integrated it.
44425 engagement10 sourcesThu, Sep 10, 2026, 12:00 AM
Developments
V4.1 Flash is now available in HuggingChat and on Hugging Face following the official September 10 launch.
Thu, Sep 10, 2026, 04:05 PM
V4.1-Flash is live on the DeepSeek API, HuggingChat, and OpenCode Go; V4 Pro requests will reroute to Flash from September 14, 2026.
Thu, Sep 10, 2026, 04:05 PM
Weights for DeepSeek-V4.1-Flash are now published on Hugging Face following the API beta.
Thu, Sep 10, 2026, 05:57 AM
AI Overview
- DeepSeek's V4.1 Flash beta uses a new architecture with native multimodal support, faster speeds, and lower costs.
- Called via model name deepseek-flash; retired names deepseek-v4-flash and deepseek-v4-flash-vision-exp temporarily route to V4.1 Flash.
- Pricing set from September 10 at $0.003 input cache hits, $0.15 cache misses, and $0.6 output off-peak, double at peak, per the DeepSeek announcement.
- DeepSeek says V4.1 Flash has comprehensively surpassed V4 Pro across performance, cost, speed, and task completion time.
- After launch, all V4 Pro requests will be routed to V4.1 Flash and billed at Flash prices until V4.1 Pro ships.
- Off-peak pricing from 12:00 Beijing time Sept 10: $0.003 per input cache hit, $0.15 per cache miss, $0.6 per output; peak rates are double.
- V4.1 Flash is the smallest model in DeepSeek's new Causal-Encoder-Decoder architecture family, a 552B-parameter MoE activating 8B parameters on input and 16B on output, with native visual understanding.
- KV cache compression cuts HBM requirements to 1/4 and SSD requirements to 1/8 versus the previous generation; cache has shrunk 437× since the first-generation model.
Entities
Related
Social·Community
Devs weigh swapping Claude for DeepSeek V4.1 Flash/Pro: cheaper tokens, mixed benchmarks, Responses API gaps
A Reddit user considering dropping Claude for cost reasons compiled a comparison of DeepSeek V4.1 Flash/Pro against Claude Sonnet 5 and Opus 5, citing benchmark scores, API pricing, and API compatibility caveats, and asked others who have switched for real-world results on tool reliability and cost per completed task.
A Reddit user considering dropping Claude for cost reasons compiled a comparison of DeepSeek V4.1 Flash/Pro against Claude Sonnet 5 and Opus 5, citing benchmark scores, API pricing, and API compatibility caveats, and asked others who have switched for real-world results on tool reliability and cost per completed task.
10 engagement1 sourceredditSun, Sep 20, 2026, 04:05 PM
Social·Community
DeepSeek V4.1 Flash API: Pricing, 1M Context & Fast Inference Guide
Medium guide covers DeepSeek V4.1 Flash API pricing, 1M context, and inference speed
Medium guide covers DeepSeek V4.1 Flash API pricing, 1M context, and inference speed
3.5k engagement2 sourcesrss, youtubeTue, Oct 6, 2026, 06:53 PM
Social·Communityfeatured
DeepSeek details V4.1 Flash architecture in new paper
Two Minute Papers examines DeepSeek V4.1 Flash's new architecture in video breakdown
Two Minute Papers examines DeepSeek V4.1 Flash's new architecture in video breakdown
4.4k engagement1 sourceyoutubeFri, Sep 18, 2026, 07:03 AM
Social·Community
Reddit user reports DeepSeek v4.1-flash burning ~4x more spend than v4-flash-0731 despite lower published pricing
A Reddit user reports DeepSeek v4.1-flash burning roughly 4x more spend than v4-flash-0731 in production, even though 4.1's published per-1M rates are lower. The user speculates the gap implies 4-5x token consumption per call or non-published billing factors, and asks whether others see the same.
A Reddit user reports DeepSeek v4.1-flash burning roughly 4x more spend than v4-flash-0731 in production, even though 4.1's published per-1M rates are lower. The user speculates the gap implies 4-5x token consumption per call or non-published billing factors, and asks whether others see the same.
23 engagement2 sourcesredditSun, Sep 13, 2026, 05:24 PM
Social·Community
Operation Cheepseek Phase 3: $80 of usage for DeepSeek V4.1 Flash
User reports $80 of usage on DeepSeek V4.1 Flash in 'Operation Cheepseek Phase 3'
User reports $80 of usage on DeepSeek V4.1 Flash in 'Operation Cheepseek Phase 3'
8.4k engagement1 sourcexMon, Oct 12, 2026, 09:14 AM