Mistral Small 3.1 24B
by Mistral AI · mistral family
24B
parameters
text-generation code-generation reasoning multilingual vision tool-use summarization
Mistral Small 3.1 24B is a multimodal model that handles text and image inputs with 128K context. It's the first model from Mistral with native vision support and delivers strong results across general tasks, coding, and multilingual work. At Q4 it needs about 18 GB VRAM — fits on a RTX 3090 or a Mac with 24 GB. Positioned as Mistral's sweet spot between efficiency and capability, it's well-suited for daily use on high-end consumer hardware.
Quick Start with Ollama
ollama run 24b-instruct-q4_K_M | Creator | Mistral AI |
| Parameters | 24B |
| Architecture | transformer-decoder |
| Context | 128K tokens |
| Released | Mar 18, 2025 |
| License | Apache 2.0 |
| Ollama | mistral-small3.1 |
Quantization Options
| Format | File Size | VRAM Required | Quality | Ollama Tag |
|---|---|---|---|---|
| Q4_K_M rec | 15 GB | 18 GB | | 24b-instruct-q4_K_M |
| Q8_0 | 26 GB | 30 GB | | 24b-instruct-q8_0 |
| F16 | 49 GB | 54 GB | | 24b-instruct-fp16 |
Compatible Hardware
Q4_K_M requires 18 GB VRAM
Benchmark Scores
78.0
mmlu