Skip to content

Qwen3-Coder vs Devstral Small 2

Qwen3-Coder and Devstral Small 2 side by side: model sizes, the VRAM you need, context window, abilities, licence and which one to pick.

Side by side

Qwen3-CoderDevstral Small 2
TypeModel in the Qwen Coder familyModel in the Devstral family
DeveloperAlibaba Cloud (Qwen team)Mistral AI
ReleasedJul 2025Dec 2025
Parameters30B (3B active), 480B (35B active)24B
Sizes30B · 480B24B
Smallest VRAM (4-bit)~21.5 GB~17 GB
Context window256K tokens384K tokens
AbilitiesTool callingTool callingImage input
LicenceApache 2.0Apache 2.0
Run itollama run qwen3-coder:30bollama run devstral-small-2:24b

Which one should you pick?

  • Devstral Small 2 runs on smaller hardware: from ~17 GB of GPU memory, against ~21.5 GB.
  • Devstral Small 2 reads longer inputs: a 384K-token context window.
  • Only Devstral Small 2 has image input.

Pick Qwen3-Coder if…

  • your tools or fine-tunes already target it (released Jul 2025, so support is mature)

Best for: Fast MoE coder that fits a 24 GB GPU

Pick Devstral Small 2 if…

  • you have less GPU memory (it starts at 17 GB)
  • you need longer inputs (384K tokens)
  • you need image input
  • you want the newer model (Dec 2025)

Best for: Coding agent on one GPU, image input, 384K context

VRAM and download size

Ollama's default 4-bit builds with an 8K context. Read GB as system RAM on a CPU-only machine or a Mac.

Qwen3-Coder

ModelDownloadVRAM
30B19 GB~21.5 GB
480B290 GB~319.5 GB

Devstral Small 2

ModelDownloadVRAM
24B15 GB~17 GB

Frequently asked questions

Is Qwen3-Coder better than Devstral Small 2?

It depends on the job and your hardware. Devstral Small 2 runs on smaller hardware: from ~17 GB of GPU memory, against ~21.5 GB. Devstral Small 2 reads longer inputs: a 384K-token context window. Only Devstral Small 2 has image input. For quality, test both on your own prompts: benchmark rankings change with every release.

Which needs less VRAM, Qwen3-Coder or Devstral Small 2?

Qwen3-Coder starts at ~21.5 GB (19 GB download). Devstral Small 2 starts at ~17 GB (15 GB download).

Can I use Qwen3-Coder and Devstral Small 2 commercially?

Qwen3-Coder: Apache 2.0. Devstral Small 2: Apache 2.0. Check the exact licence of the model you deploy; custom licences can limit very large companies or certain uses.

More comparisons

← All AI models

Back to top