Alex Getmanearly AI signals
EN

DeepSeek V4 Flash

DeepSeek

ReleasedUpdated
Context window
1M
1,048,576 tokens
Input / Output
text → text

Timeline

  1. · Weights version · evidence link
  1. · Evidence links · GitHub search · DeepSeek · commits · rgolde96/natively-ai-personal · @unknown
    Excerpt
    feat(llm): DeepSeek V4 Flash on OpenRouter with reasoning off and Fireworks/Wafer/DeepInfra order (D-30); record owner answers, WS23, owner info page
  2. · Evidence links · GitHub search · DeepSeek · commits · nikgraf/calendar · @nikgraf
  3. · Evidence links · GitHub search · DeepSeek · commits · miron404/lindwyrm · @miron404
    Excerpt
    defaulted to the retired deepseek-v4-flash model name.
  4. · Evidence links · GitHub search · DeepSeek · commits · genji1024/dotfiles · @genji1024
  5. · Evidence links · GitHub search · DeepSeek · commits · bottomless/agent-duel · @shihabcodes
  6. · Evidence links · GitHub search · DeepSeek · commits · ModernNomad-98/Project-Aegis · @deepseek-v4-flash
  7. · Evidence links · GitHub search · DeepSeek · commits · ModernNomad-98/Project-Aegis · @deepseek-v4-flash
  8. · Evidence links · GitHub search · DeepSeek · pulls · overnightworks/speech · @dependabot[bot]
  9. · Evidence links · GitHub search · DeepSeek · pulls · nikgraf/calendar · @nikgraf
  10. · Evidence links · GitHub search · DeepSeek · pulls · maximhq/bifrost · @jimaldon
    Excerpt
    needed. The model catalog datasheet already carries `tencent/deepseek-v4-pro`, `tencent/deepseek-v4-flash` and `tencent/minimax-m3`; live-only models are listed with blank pricing and can be priced via overrides.
  11. · Evidence links · GitHub search · DeepSeek · pulls · maximhq/bifrost · @Amit-kr26
    Excerpt
    [x] Live run with a real key against `deepseek-v4-flash`: simple chat, multi-turn, streaming, tool calls (single/multiple/automatic/end-to-end), list models (incl. pagination), Responses API (incl. streaming), and
  12. · Evidence links · GitHub search · DeepSeek · pulls · maximhq/bifrost · @neta79
    Excerpt
    -d '{"model":"deepseek-v4-flash","messages":[{"role":"user","content":"Say exactly: hello world"}],"max_tokens":20}'
  13. · Evidence links · GitHub search · DeepSeek · pulls · maximhq/bifrost · @neta79
    Excerpt
    "model": "deepseek/deepseek-v4-flash",
  14. · Evidence links · GitHub search · DeepSeek · pulls · gufo-org/gufo · @mduka
  15. · Evidence links · GitHub search · DeepSeek · pulls · anomalyco/models.dev · @models-dev-automation[bot]
  16. · Evidence links · GitHub search · DeepSeek · pulls · alibaba/rtp-llm · @fxp19950422
    Excerpt
    Brings the DeepSeek V4 Flash PPU decode path to the execution shape measured in #1434. Every new behavior is opt-in or keeps its previous default.
  17. · Evidence links · GitHub search · DeepSeek · pulls · SreeSharvesh/loc-arena · @ryanlail
    Excerpt
    - `…control-note.glm-flash.yaml` and `…control-note.deepseek-v4-flash.yaml`: the same run with GLM 5.3 Flash and DeepSeek V4 Flash as the untrusted agent, each with its model file.
  18. · Evidence links · GitHub search · DeepSeek · pulls · Picazo333/nexova-ai-engineering · @Picazo333
    Excerpt
    Modelo de generación: `online/openrouter/deepseek/deepseek-v4-flash` en `https://llm.4geeks.ai/v1`. Embeddings: `online/openrouter/perplexity/pplx-embed-v1-0.6b`.
  19. · Evidence links · GitHub search · DeepSeek · pulls · NixOS/nixpkgs · @asosnovsky
  20. · Evidence links · GitHub search · DeepSeek · pulls · NeuroLang/NeuroLang · @demianw
  21. · Evidence links · GitHub search · DeepSeek · pulls · NeuroLang/NeuroLang · @demianw
  22. · Evidence links · GitHub search · DeepSeek · pulls · I-GUIDE/iguide-ai · @ShanmukhUpad
  23. · Evidence links · GitHub search · DeepSeek · pulls · BerriAI/litellm · @cloudgpuapp
    Excerpt
    Usage: `litellm.completion(model="cloudgpu/deepseek-v4-flash", messages=[...])`
  24. · Evidence links · GitHub search · DeepSeek · pulls · 0xPlaygrounds/rig · @gold-silver-copper
  25. · Evidence links · DeepSeek · updates
  26. · Alex's posts · Alex Getman
    Excerpt
    🚨 DeepSeek slashed OpenCode Go limits 17x DeepSeek V4 Flash limits dropped from 63,300 to 3,800, while V4 Pro fell from 3,450 to 1,050. This comes as DeepSeek API prices have jumped sharply.
  27. · Alex's posts · Alex Getman
    Excerpt
    ⚡️ Alibaba released Qwen 3.8 Max The new 2.4T parameter flagship benchmarks in the same league as GPT-5.6 Sol, Claude Opus 4.8 and Fable 5. API pricing is $2/M input and $6/M output, roughly half the cost of GPT-5.6 Terra and Claude Sonnet. Now it’s up to users to see whether Qwen can beat DeepSeek-V4-Flash and GPT-5.6 Luna on price to performance.
  28. · Alex's posts · Alex Getman
    Excerpt
    Yesterday GPT 5.6 Luna pushed the Pareto frontier for intelligence and cost. Today DeepSeek V4 Flash did it again and pushed it even further.
Specifications

DeepSeek V4 Flash is a model from DeepSeek with open weights, first recorded on 20 Sep 2026; its release date is not established. The context window is 1,048,576 tokens.

Context window
1,048,576
Input
text
Output
text
Weights
Open

Questions

What is the context window of DeepSeek V4 Flash?

The context window is 1,048,576 tokens.

Are the weights of DeepSeek V4 Flash publicly available?

Yes. DeepSeek publishes the weights.

When was DeepSeek V4 Flash released?

Its release date is not established. Signal Forge first recorded it on 20 Sep 2026.