Mistral Small 24B
A 24 billion-parameterHow many numbers the model learned during training — the usual rough measure of its size. dense model from Mistral AI. A latency-optimised 24B model.
- Size
- 23.6Bdense
- Memory to run
- ~15 GBsmallest, 8K ctx
- Context
- 32Ktokens
- License
- PermissiveApache License 2.0
- Updated
- 13 Sept 2026
Which version to use
- Variant Mistral Small 24B Instruct 2501 · Mistral AITuned to follow instructions and hold a conversation. The usual choice.
- Variant Dolphin Mistral 24B Venice Edition · rbinrsA community or third-party fine-tune of another variant.
Where it runs
Best-fitting version on each reference system, at an 8K contextHow much text the model can consider at once, counted in tokens — roughly ¾ of a word each..
Runs well 11
- A100 80GB server~28 tok/s est.
- Dual RTX 3090~29 tok/s est.
- GB10 mini workstation~4 tok/s est.
- MacBook Pro M3 Max 64 GB~13 tok/s est.
- MacBook Pro M4 Max 128 GB~18 tok/s est.
- Mac mini M4 Pro 48 GB~9 tok/s est.
- Mac Studio M2 Ultra 192 GB~26 tok/s est.
- RTX 4090 workstation~33 tok/s est.
- RTX 5090 workstation~59 tok/s est.
- RX 7900 XTX desktop~31 tok/s est.
- Ryzen AI Max+ 395 mini PC~4 tok/s est.
Slowly (CPU or offload) 2
- CPU-only Ryzen 9 7950X~3 tok/s est.
- RTX 4060 Ti 16GB budget build~12 tok/s est.
Too large 0
None of the reference systems.
Benchmarks
Each benchmarkA fixed set of questions every model is given, so their scores can be compared on the same task. is developer-reported; prompts and settings differ between labs. Bars are relative to the best open result.
Variants & downloads
Each variant is a separate set of weights. Expand one for its downloads: quantizationStoring each of the model’s numbers with fewer bits, so the file is smaller and needs less memory. trades a little quality for a much smaller file, and the memory column adds the working memoryExtra space the model needs while it answers. It grows with the length of the conversation, on top of the file itself. a conversation needs on top.
Variant Mistral Small 24B Instruct 2501Permissive
- What it is
- Tuned to follow instructions and hold a conversation. The usual choice.
- Publisher
- Mistral AI
- License
- Apache License 2.0 · commercial use allowed
- Released
- 30 Jan 2025
- huggingface
- mistralai/Mistral-Small-24B-Instruct-2501
Contributions are not open yet, so there is nothing here from members.
| Quantization | Format | Bits | Download | Memory @ 8K | Publisher |
|---|---|---|---|---|---|
| Download BF16native | safetensors | 16 | 43.9 GiB | Mistral AImistralai/Mistral-Small-24B-Instruct-2501 | |
| Download Q6_Kk quant | gguf | 6.56 | 18.0 GiB | LM Studio Communitylmstudio-community/Mistral-Small-24B-Instruct-2501-GGUF | |
| Download Q4_K_Mk quant | gguf | 4.89 | 13.4 GiB | LM Studio Communitylmstudio-community/Mistral-Small-24B-Instruct-2501-GGUF | |
| Download MLX 4-bitmlx | mlx | 4.5 | 12.3 GiB | MLX Communitymlx-community/Mistral-Small-24B-Instruct-2501-4bit |
Variant Dolphin Mistral 24B Venice EditionPermissive
- What it is
- A community or third-party fine-tune of another variant.
- Publisher
- rbinrs
- License
- Apache License 2.0 · commercial use allowed
- Released
- 12 Sept 2026
- huggingface
- rbinrs/Dolphin-Mistral-24B-Venice-Edition
Contributions are not open yet, so there is nothing here from members.
No downloadable artifacts recorded.
Architecture details
- Parameters
- 23.57B
- Active / token
- All (dense)
- Architecture
- Dense
- Layers
- 40
- Attention heads
- 32
- KV heads
- 8
- Head dim
- 128
- KV cache @ 8K (fp16)
- 1.25 GiB
- Max context
- 32,768 tokens
Measured performance
Throughput on specific systems and runtimes, with the source of each measurement.
Community results
Reviews
Lineage
How this model’s variants relate to each other and to other models.
Sources & history
Sources
- Hugging Face Hub (live source, 1 record, 13 Sept 2026)
- Mutinai illustrative fixtures (illustrative fixture, 1 record, 13 Sept 2026)
External identifiers
- huggingface mistralai/Mistral-Small-24B-Instruct-2501
- huggingface rbinrs/Dolphin-Mistral-24B-Venice-Edition
Field history
- base = "mistralai/Mistral-Small-24B-Instruct-2501" — Hugging Face Hub (current)
- derivation = "fine_tune" — Hugging Face Hub (current)
- license = "apache-2.0" — Hugging Face Hub (current)
- name = "Dolphin Mistral 24B Venice Edition" — Hugging Face Hub (current)
Timeline
- Mistral Small 3 (24B) released under Apache 2.0