ZeroHour
Story · 2 sources · 3 articlesfirst updated ()1

DeepSeek launches V4.1 Flash with claims of beating V4 Pro; uncensored third-party FP8 build appears on Hugging Face

infoModel releaseimportance 55
What's new: Since the previous summary (2026-09-10T07:22Z), a third-party uncensored FP8 quantization of DeepSeek v4.1 Flash by Hugging Face user dealignai has appeared (2026-09-11), the first publicly reported model artifact following the terse official announcement. The launch window (~September 10, 2026) has passed with no official benchmarks, parameter counts, capability details, or V4.1 Pro release date…
Merged summary · glm-5.3-flash · rewritten as coverage arrives

DeepSeek announced V4.1 Flash via its official X account around September 10, 2026, claiming it surpasses V4 Pro on performance, cost, speed, and task completion time, with Pro traffic rerouted to Flash at Flash pricing until V4.1 Pro ships. The announcement…

DeepSeek announced V4.1 Flash, launching around September 10, 2026, via a terse post on its official X account. The company claims V4.1 Flash surpasses V4 Pro across performance, cost, speed, and task completion time, but the announcement included no benchmarks, parameter counts, or capability details, and no further official information was available at publication, leaving the claimed superiority over V4 Pro unsubstantiated by published data. After launch and before V4.1 Pro ships, all Pro requests will be routed to Flash and billed at Flash pricing. Off-peak prices are $0.003 for input cache hits, $0.15 for cache misses, and $0.60 for output, with peak-hour rates doubled. As of the latest report (2026-09-11), an unofficial community release has appeared: Hugging Face user dealignai published an uncensored FP8-quantized variant of DeepSeek v4.1 Flash. This is a third-party upload rather than an official DeepSeek release, provides no benchmark data or license details, and received limited visibility (43 points and 11 comments on Hacker News). No official technical details, benchmarks, or V4.1 Pro release date have been reported.

  • DeepSeek announced V4.1 Flash via its official X account, with launch set around September 10, 2026 (Report 1: 2026-09-09; Report 2: 2026-09-10).
  • DeepSeek claims V4.1 Flash surpasses V4 Pro across performance, cost, speed, and task completion time; the announcement contained no benchmarks, parameter counts, or capability details, so the claim is not backed by published data.
  • After launch and before V4.1 Pro ships, all Pro requests will be routed to Flash and billed at Flash pricing.
  • Off-peak pricing: $0.003 for input cache hits, $0.15 for cache misses, $0.60 for output; peak-hour prices are double the off-peak rates.
  • On 2026-09-11, Hugging Face user dealignai published an uncensored FP8 quantization of DeepSeek v4.1 Flash; it is a third-party upload, not an official DeepSeek release, with no benchmark data or license details in the listing.
  • The community quantization received 43 points and 11 comments on Hacker News, described as limited visibility.
  • No official V4.1 Pro release date, benchmarks, or parameter counts have been reported as of 2026-09-11.

Coverage timeline

  1. · 6d ago
    Hacker News · security· 55
    DeepSeek launching v4.1 flash cheaper and more capable than v4 pro

    DeepSeek will launch V4.1 Flash around September 10, 2026, claiming it outperforms V4 Pro on performance, cost, and speed while routing Pro traffic to it.

  2. · 5d ago
    Hacker News · AI· 50
    DeepSeek v4.1 Flash

    DeepSeek announced v4.1 Flash, a new model version, in a terse social media post without technical details.

  3. · 4d ago
    Hacker News · AI· 30
    DeepSeek v4.1 Flash Uncensored

    Hugging Face user dealignai published an uncensored FP8-quantized variant of DeepSeek v4.1 Flash, drawing moderate Hacker News attention.