DeepSeek V4 Flash

DeepSeek AI, open modeldeepseek-v4-flashModel source: Hugging Face

Fast, with a huge memory: it reads a whole contract without losing the thread.

Reads in one go

≈ 1,100 pages

524,288 tokens

Brain size

The largest in the catalogue

1.6T MoE, FP8

Where it runs

Spain (EU)

2× B200

Understands

Text, not images

Chat, Reasoning, Tools

What it is good at

Reasoning. Thinks in several steps before answering, for problems that take more than one.
Tools. Searches, queries data and calls other systems. The basis of an agent.
MoE. Mixture of Experts: it splits big tasks and shares them among specialists, switching on only the ones it needs. It performs like a huge model without costing like one.
Long context. Reads enormous documents in one go without losing the thread.
Chat. Converses, summarises and writes in plain language.

Only need simple chat, classification or translation? Qwen3.5 9B does it for a fraction of the price, and switching model is switching one word.

Pricing

€ per million tokens, VAT included

You send (input)0.14 €

It answers (output)0.28 €

Re-reading repeats (cache)0.014 €

1,000 chat answers~0.15 €

Analysing a 200 page contract~0.01 €

try it with 50 € freeActivated with your card. Enough for about 324,675 answers.

For your technical team

Compatible with the OpenAI SDK: change the base URL and the key, and that is it.

curl https://api.gpuflow.ai/v1/chat/completions \
  -H "Authorization: Bearer $GPUFLOW_API_KEY" \
  -d '{ "model": "deepseek-v4-flash",
        "messages": [{ "role": "user", "content": "Hola" }] }'