作者： Jakub Rusinowski · 最后更新： 2026年9月6日
Mistral's reasoning line: Mistral Small with long chain-of-thought added via SFT on Magistral Medium traces plus RL. Apache 2.0, vision-capable, and sized so the whole thing fits a single RTX 4090 or a 32 GB MacBook once quantized — which is the point. Reasoning traces degrade past roughly 40K tokens even though the context window runs to 128K.
| Licence | What it permits | Applies to |
|---|---|---|
Apache-2.0 | Commercial use permitted Commercial use permitted. No usage restrictions beyond attribution. | Magistral Small 24B |
| Magistral Small 24B | Min 15 GB VRAM · Q4_K_M · 131,072 ctx · ollama run magistral:24b |
The cheapest GPU that runs Magistral Small locally (min 15 GB VRAM) is the AMD Radeon RX 9060 XT 16GB (16 GB).
Install Ollama then run: ollama run magistral:24b
Minimum VRAM: 15 GB. For best results use Q4_K_M quantization.
Magistral Small needs about 15 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: Magistral Small 24B (15 GB, Q4_K_M). On Apple Silicon, unified memory counts toward this requirement.
Yes — Magistral Small runs on an RTX 4090 (24 GB) and other 24 GB cards such as the RTX 3090. Smaller variants also fit comfortably on 8–16 GB GPUs at Q4_K_M.
Q4_K_M is the best balance of quality and VRAM for Magistral Small in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
Install Ollama, then run: ollama run magistral:24b. This downloads Magistral Small and starts a local, OpenAI-compatible endpoint — no internet connection is needed after the initial download.