Devstral Small 2505 24B — VRAM, speed & local setup
Written by Jakub Rusinowski · Last updated
The first-generation Devstral Small (release 2505), and still the local coding model with the clearest published number attached to it: 46.8% on SWE-Bench Verified, which beat the prior open-source state of the art by 6 points at release. Finetuned from Mistral Small 3.1, so it inherits the 128K context. At 24B dense it is light enough for a single RTX 4090 or a 32GB Mac (~14 GB at Q4_K_M). Apache 2.0.
Devstral Small 2505 24B needs about 15 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
VRAM and speed by quantization
Quoted against NVIDIA RTX 4090 (24 GB). Includes the KV cache at 8K context, so it reads higher than the headline figure.
| Quant | Memory | VRAM | Speed (est.) | Fit |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 10 GB | ~74 tok/s | Fits | |
| Q3_K_M 3.41 bpw | 12.4 GB | ~60 tok/s | Fits | |
| Q4_K_M 4.83 bpw | 16.6 GB | ~45 tok/s | Fits | |
| Q5_K_M 5.67 bpw | 19.2 GB | ~39 tok/s | Fits | |
| Q6_K 6.56 bpw | 21.8 GB | ~34 tok/s | Tight | |
| Q8_0 8.50 bpw | 27.6 GB | ~4 tok/s | Offload | |
| F16 16.00 bpw | 50.1 GB | ~2 tok/s | Offload |
Black marker = usable memory on the NVIDIA RTX 4090 (24 GB). Estimates from the memory-bandwidth roofline on the methodology page. Devstral Small 2505 24B VRAM calculator →
Get Devstral Small 2505 24B running
The cheapest catalogued GPU that runs Devstral Small 2505 24B is the AMD Radeon RX 7900 XT (20 GB).
or compare on Vast.ai from $0.35/hr (typical low · varies)
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
How to run Devstral Small 2505 24B
Install Ollama, then run:
ollama run devstral:24bSpecifications
Preview — The model is released, but these specs are thin or rest on a single source. Individual fields may be wrong.
Preview — The model is released, but these specs are thin or rest on a single source. Individual fields may be wrong.
- Parameters
- 24 Billion
- Context window
- 128,000
- Architecture
- Dense Transformer
- Provider
- Mistral AI
- Licence
- Apache 2.0
- Practical deployment
- Single consumer GPU
- Specified at
- Q4_K_M
- System RAM
- 32 GB
- Record updated
- 2026-08-15
Commercial use permitted. No usage restrictions beyond attribution.
Quality and use cases
Scores as published by the model’s authors or an independent evaluator — quality, not throughput, and not measured by us.
| Benchmark | Score | Provenance |
|---|---|---|
| SWE-Bench Verified | 46.8 / 100 % | reported · https://mistral.ai/news/devstral/ |
Can I run Devstral Small 2505 24B on my GPU?
- Devstral Small 2505 24B on AMD Radeon RX 7800 XT
- Devstral Small 2505 24B on AMD Radeon RX 7900 XT
- Devstral Small 2505 24B on AMD Radeon RX 7900 XTX
- Devstral Small 2505 24B on AMD Radeon RX 9060 XT 16GB
- Devstral Small 2505 24B on AMD Radeon RX 9060 XT 8GB
- Devstral Small 2505 24B on AMD Radeon RX 9070
- Devstral Small 2505 24B on AMD Radeon RX 9070 XT
- Devstral Small 2505 24B on AMD Ryzen AI Max+ 395
- Devstral Small 2505 24B on Apple M1
- Devstral Small 2505 24B on Apple M1 Pro
- Devstral Small 2505 24B on Apple M1 Ultra
- Devstral Small 2505 24B on Apple M2
- Devstral Small 2505 24B on Apple M2 Pro
- Devstral Small 2505 24B on Apple M3
- Devstral Small 2505 24B on Apple M3 Pro
- Devstral Small 2505 24B on Apple M4
- Devstral Small 2505 24B on Apple M4 Max
- Devstral Small 2505 24B on Apple M4 Pro
- Devstral Small 2505 24B on Apple M5
- Devstral Small 2505 24B on Apple M5 Max
- Devstral Small 2505 24B on Intel Arc B570
- Devstral Small 2505 24B on Intel Arc B580
- Devstral Small 2505 24B on NVIDIA A100 80GB (PCIe)
- Devstral Small 2505 24B on NVIDIA DGX Spark
- Devstral Small 2505 24B on NVIDIA GeForce RTX 3060 (12GB)
- Devstral Small 2505 24B on NVIDIA GeForce RTX 3070
- Devstral Small 2505 24B on NVIDIA GeForce RTX 3070 Ti
- Devstral Small 2505 24B on NVIDIA GeForce RTX 3080 (10GB)
- Devstral Small 2505 24B on NVIDIA GeForce RTX 3090
- Devstral Small 2505 24B on NVIDIA GeForce RTX 4060
- Devstral Small 2505 24B on NVIDIA GeForce RTX 4060 Ti 16GB
- Devstral Small 2505 24B on NVIDIA GeForce RTX 4070
- Devstral Small 2505 24B on NVIDIA GeForce RTX 4070 Super
- Devstral Small 2505 24B on NVIDIA GeForce RTX 4070 Ti
- Devstral Small 2505 24B on NVIDIA GeForce RTX 4070 Ti Super
- Devstral Small 2505 24B on NVIDIA GeForce RTX 4080
- Devstral Small 2505 24B on NVIDIA GeForce RTX 4080 Super
- Devstral Small 2505 24B on NVIDIA GeForce RTX 4090
- Devstral Small 2505 24B on NVIDIA GeForce RTX 5060
- Devstral Small 2505 24B on NVIDIA GeForce RTX 5060 Ti 16GB
- Devstral Small 2505 24B on NVIDIA GeForce RTX 5060 Ti 8GB
- Devstral Small 2505 24B on NVIDIA GeForce RTX 5070
- Devstral Small 2505 24B on NVIDIA GeForce RTX 5070 Ti
- Devstral Small 2505 24B on NVIDIA GeForce RTX 5080
- Devstral Small 2505 24B on NVIDIA GeForce RTX 5090
- Devstral Small 2505 24B on NVIDIA H100 80GB (PCIe)
- Devstral Small 2505 24B on NVIDIA RTX PRO 6000 Blackwell
Other Devstral sizes
Devstral Small 2505 24B — frequently asked questions
How much VRAM does Devstral Small 2505 24B need?
About 15 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does Devstral Small 2505 24B run on an RTX 4090 (24 GB)?
Yes. Devstral Small 2505 24B needs about 15 GB at Q4_K_M, inside a 24 GB card, at an estimated 45 tokens/sec.
How do I run Devstral Small 2505 24B locally?
Install Ollama and run `ollama run devstral:24b`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.
What other sizes does Devstral come in?
Devstral-2 123B (75 GB), Devstral 2 22B (14 GB), Devstral Small 2505 24B (15 GB), Devstral Small 2 24B (15 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.