UTC

DeepSeek V4 Flash

DeepSeek · open weights · served by 72 providers · tool calling, reasoning

deepseek/deepseek-v4-flash

Weights
Open
Released
2026-04-24
Knowledge cutoff
2025-05
Context window
1M
Deprecation dates

When each platform deprecates DeepSeek V4 Flash, as LiteLLM records it. Plan the move before the date: after it, the platform may stop serving the model or stop updating it.

  • wandbdeprecated 2026-10-05wandb/deepseek-ai/DeepSeek-V4-Flashsuccessor: wandb/deepseek-ai/DeepSeek-V4-Flash-0731
Every upcoming deprecation →
Hugging Face repository

deepseek-ai/DeepSeek-V4-Flash at commit 60d8d70770, last changed 2026-06-22

Licence
mit
Access
Open download
Library
transformers
Formats
.safetensors
  • No model signature file is published with the weights.

What the Hub's scanners report

22 weight files, 22 with a scan result. No scanner reports an issue with them.

The model at this commit, its 22 weight files with their SHA-256.

Facts from the Hugging Face Hub and its file scanners (Protect AI, Cisco, Hugging Face picklescan, VirusTotal, JFrog), as of 2026-11-18. ShieldMap does not copy model cards.

Who serves it
  • Abacusdeepseek-ai/DeepSeek-V4-Flashdeprecated
  • AgentRouterdeepseek-v4-flash
  • AIHubMixalicloud-deepseek-v4-flashdeprecated
  • AIHubMixdeep-deepseek-v4-flashdeprecated
  • AI/ML APIdeepseek/deepseek-chat
  • AI/ML APIdeepseek/deepseek-v4-flash
  • Alibaba (China)deepseek-v4-flashdeprecated
  • Alibaba Token Plandeepseek-v4-flashdeprecated
  • Alibaba Token Plan (China)deepseek-v4-flashdeprecated
  • AnyAPIdeepseek/deepseek-v4-flash
  • Aurikodeepseek-v4-flash
  • Azuredeepseek-v4-flash
  • Azure Cognitive Servicesdeepseek-v4-flash
  • BytePlus ModelArkdeepseek-v4-flash-260425
  • ClinePasscline-pass/deepseek-v4-flashdeprecated
  • CommandCodedeepseek/deepseek-v4-flash
  • CommandCodedeepseek/deepseek-v4-flash-fast
  • Deep Infradeepseek-ai/DeepSeek-V4-Flash
  • DigitalOceandeepseek-4-flash
  • EBCloudDeepSeek-V4-Flash
  • Eden AIdeepseek/deepseek-v4-flash
  • GMI Clouddeepseek-ai/DeepSeek-V4-Flash
  • HPC-AIdeepseek/deepseek-v4-flash
  • Huawei Cloud MAASdeepseek-v4-flash
48 more providers

Hugging Face (deepseek-ai/DeepSeek-V4-Flash), Impossibl (deepseek/deepseek-v4-flash), Inference APIs (deepseek-ai/DeepSeek-V4-Flash), InferX (deepseek-v4-flash), Jalapeno Cloud (DeepSeek-V4-Flash), JZSTOKEN (deepseek-v4-flash), Kenari (deepseek-v4-flash), Kilo Gateway (deepseek/deepseek-v4-flash), DevPass (LLM Gateway) (deepseek-v4-flash), LLM Gateway (consensusprotocol/deepseek-v4-flash), LLM Gateway (deepinfra/deepseek-v4-flash), LLM Gateway (flexai/deepseek-v4-flash), LLM Gateway (inference.net/deepseek-v4-flash), LLM Gateway (runware/deepseek-v4-flash), Merge Gateway (deepseek/deepseek-v4-flash), Moark (DeepSeek-V4-Flash), Modelis (deepseek-v4-flash), Moneo (deepseek-v4-flash:web), NanoGPT (deepseek/deepseek-v4-flash), NanoGPT (deepseek/deepseek-v4-flash:thinking), Obscurify.ai (deepseek/deepseek-v4-flash), Ofox (deepseek/deepseek-v4-flash), OpenRouter (deepseek/deepseek-v4-flash), Opper (deepseek-v4-flash), OrcaRouter (deepseek/deepseek-v4-flash), OrcaRouter (deepseek/deepseek-v4-flash-free), Pendra (deepseek-v4-flash), Pioneer (deepseek-ai/DeepSeek-V4-Flash), Poe (empiriolabs/deepseek-v4-flash-el), Prime Inference (deepseek/deepseek-v4-flash), Requesty (deepseek-v4-flash), routing.run (deepseek-v4-flash), SandBase (deepseek/deepseek-v4-flash), SCNet Token Plan (DeepSeek-V4-Flash), SCX.ai (DeepSeek-V4-flash), SenseNova (China) (deepseek-v4-flash), SiliconFlow (deepseek-ai/DeepSeek-V4-Flash), SiliconFlow (China) (deepseek-ai/DeepSeek-V4-Flash), Standard Compute (deepseek/deepseek-v4-flash), SteadyRouter (deepseek-v4-flash), TokenGo (deepseek/deepseek-v4-flash), UnoRouter (deepseek-v4-flash), UnoRouter (deepseek-v4-flash:free), Venice AI (deepseek-v4-flash), Vercel AI Gateway (deepseek/deepseek-v4-flash), Volcengine Ark Coding Plan (deepseek-v4-flash), Watt Cloud (DeepSeek-V4-Flash), ZenMux (deepseek/deepseek-v4-flash)

Running it yourself

Open weights are loaded and served by code that has its own advisories. These inference servers and model loaders are on ShieldMap; load weights only from the publisher's own repository, and prefer safetensors over formats that can run code.

How supply-chain attacks on models work →

Sources: models.dev (MIT) and the LiteLLM model catalogue (MIT). Deprecation dates are as each platform published them and as LiteLLM records them; check the platform before you rely on one. Data as of 2026-11-22.