Can Desktop PC · 2× GeForce RTX 3090 (48 GB) · 128 GB RAM run Devstral Small 2 (24B)?
48 GB VRAM + 128 GB RAM · 936 GB/s memory bandwidth · Devstral Small 2 (24B) has 24B parameters.
Short answer
Yes, excellent for chat and comfortable or better for every workload we test.
Best quantization for chat: Q5_K_M (16 GB of weights) on Ollama. Generation: 25–45 tok/s, leaving about 29 GB free for your other apps.
How it rates for each workload
Same model and machine, different jobs. Agents and long documents send far bigger prompts than chat, so they need much more speed and memory.
| Workload | Verdict | Quant | Generation | First reply | Context | |
|---|---|---|---|---|---|---|
| Chatwith Open WebUI | Excellent | Q5_K_M | 25–45 tok/s | under 1 s | 8K | Details |
| Coding questionswith Continue | Excellent | Q6_K | 20–35 tok/s | ≈3 s | 16K | Details |
| Coding in a repositorywith Aider | Comfortable | Q6_K | 20–35 tok/s | ≈9 s | 32K | Details |
| Agentic codingwith OpenCode | Comfortable | Q5_K_M | 20–35 tok/s | ≈10 s | 32K | Details |
| Long documents (64K)with Open WebUI | Excellent | Q6_K | 16–25 tok/s | ≈45 s | 64K | Details |
Speeds are bandwidth-based estimates; no direct benchmark exists for this pair. “First reply” is the time to read the workload’s opening prompt. How we calculate
Get it running
Copy-paste commands for the Q6_K build with a 64K context. Ollama and llama.cpp download the file straight from Hugging Face.
ollama run hf.co/unsloth/Devstral-Small-2-24B-Instruct-2512-GGUF:Q6_K
Downloads 19.35 GB (Q6_K). Ollama's default context is small: set OLLAMA_CONTEXT_LENGTH=65536 before `ollama serve` for long prompts.
Other models for Desktop PC · 2× GeForce RTX 3090 (48 GB) · 128 GB RAM
The best-rated models for chat on this machine.
- Qwen3.6 35B-A3B140–230 tok/sExcellent
- Gemma 4 26B-A4B110–180 tok/sExcellent
- Qwen3.5 35B-A3B140–230 tok/sExcellent
- LFM2.5 24B-A2B270–450 tok/sExcellent
- gpt-oss-20b160–270 tok/sExcellent
- GLM-4.7-Flash (30B-A3B)130–220 tok/sExcellent
Cheapest machines for Devstral Small 2 (24B)
Hardware rated comfortable or better for chat, lowest price first.
- Mac mini M4 Pro 24 GB$1,399 · 12–16 tok/sComfortable
- Desktop PC · Intel Arc Pro B60 24GB · 32 GB RAM$1,500 · 10–17 tok/sComfortable
- Desktop PC · Radeon RX 7900 XT · 32 GB RAM$1,550 · 30–45 tok/sExcellent
- Ryzen AI Max+ 395 mini PC / Framework Desktop · 64 GB$1,600 · 11–15 tok/sComfortable
- Desktop PC · Intel Arc Pro B60 24GB · 64 GB RAM$1,628 · 10–17 tok/sComfortable
- Mac mini M5 Pro 24 GB$1,699 · 11–18 tok/sComfortable
Different tool, context size or workload?
These pages use typical settings. The full check models your exact setup.