← Every modelAlibaba (Qwen) · 4B, dense

Qwen3 4B

The smallest Qwen3 worth running. It has the same thinking mode as its bigger siblings, so it can reason step by step when you ask, and it fits on almost anything, including laptops with no graphics card.

ollama run qwen3:4b

Needs about 4 GB of graphics memory, or unified memory on a Mac. First time? Install Ollama, then paste that into a terminal.

Download2.5 GB
Memory to runAbout 4 GB
Read per token2.5 GB
LicenseApache 2.0
Context32K tokens
ReleasedApril 2025
Good for
  • Fast chat on old or low-memory machines
  • Summarizing and rewriting short text
  • Classifying and extracting data in scripts
  • A draft model for speculative decoding
Not the best pick for
  • Long, careful coding work
  • Facts it has to remember rather than look up
146 GPUs and Macs

What runs Qwen3 4B

Estimated speed on each card at the smallest memory size that fits it, from memory bandwidth. As a dense model it reads all its weights for every token, so memory bandwidth sets the speed.

Runs great 137 cards

Usable 9 cards

Cheapest build from our list that runs it well: The computer you already have, from $0 used, at about 22 tok/s. See the builds.

Learn

Read more about Qwen3 4B

Compare

Other models