Can I Run Devstral on 192 GB system RAM?
Written by Jakub Rusinowski · Last updated August 15, 2026
Technically yes, but not recommended
It loads, but it is not worth running — Devstral-2 123B at Q8_0 fits in 192 GB system RAM's 153.6 GB, yet the memory bandwidth limits it to ~1.5 tok/s (estimated), well below usable interactive speed.
Confidence: medium · Recommended quantization: Q8_0 · Estimated speed: ~1.5 tok/s
192 GB system RAM — what it gives a model
| Usable memory for models | 153.6 GB |
| Memory bandwidth | 90 GB/s |
Devstral on 192 GB system RAM: memory by quantization
| Quant | Memory needed | Fits 153.6 GB? | Max context | Est. speed | Download |
|---|
| F16 | 249.7 GB | ✗ No | — | — | 246 GB |
| Q8_0 | 134.4 GB | ✓ Yes | 32K | ~1.5 tok/s | 130.7 GB |
| Q6_K | 104.5 GB | ✓ Yes | 64K | ~2 tok/s | 100.9 GB |
| Q5_K_M | 90.9 GB | ✓ Yes | 64K | ~2.2 tok/s | 87.2 GB |
| Q4_K_M | 77.9 GB | ✓ Yes | 64K | ~2.6 tok/s | 74.3 GB |
| Q3_K_M | 56.1 GB | ✓ Yes | 64K | ~3.6 tok/s | 52.4 GB |
| Q2_K | 44.1 GB | ✓ Yes | 64K | ~4.6 tok/s | 40.4 GB |
Which Devstral sizes fit
| Variant | Needs at Q4_K_M | Fits? | Est. speed |
|---|
| Devstral-2 123B | 77.9 GB | ✓ Fits | ~2.6 tok/s |
| Devstral Small 24B | 16.6 GB | ✓ Fits | ~4.4 tok/s |
| Devstral-2 22B | 15.7 GB | ✓ Fits | ~4.7 tok/s |
What to watch out for
- At ~1.5 tok/s this loads but is too slow for interactive use — expect roughly 40 seconds per 60 tokens.
- This model does not publish its full attention configuration, so the KV-cache share of these figures is inferred from its parameter count rather than computed exactly.
- These figures assume CPU-only inference. Any discrete GPU, even an 8 GB one, will be several times faster for models that fit in its VRAM.
Recommended setup
llama.cpp (CPU build) or Ollama — both run without a GPU
How these numbers are calculated
- Memory need = quantized weights + KV cache at 8,192 tokens (f16) + 0.8 GB runtime overhead.
- 153.6 GB of the 192 GB is treated as usable for model weights (80% — the rest is the OS and running applications).
- DDR5-5600 dual channel at 89.6 GB/s peak. CPU decode is assumed to sustain 35% of that peak, because CPU inference is not purely bandwidth-bound — it also spends real time in compute and thread synchronisation. This figure is an assumption, not a fitted constant: no CPU measurement is in the calibration set.
- CPU-only inference: no GPU is assumed. A GPU of any size will beat these figures substantially.
- Throughput is a memory-bandwidth roofline estimate, not a measurement. It is labelled "estimated" everywhere it appears.
- KV cache is inferred from the parameter count because this model does not publish its full attention configuration.
FAQ
Can I run Devstral on 192 GB system RAM?
It loads, but it is not worth running — Devstral-2 123B at Q8_0 fits in 192 GB system RAM's 153.6 GB, yet the memory bandwidth limits it to ~1.5 tok/s (estimated), well below usable interactive speed.
Which quantization of Devstral should I use on 192 GB system RAM?
Q8_0 — it needs about 134.4 GB of the 153.6 GB available, downloads as roughly 130.7 GB, and runs at an estimated 1.5 tokens/sec with up to 32K of context.
What limits Devstral on 192 GB system RAM?
Memory bandwidth. The model fits, but at 89.6 GB/s it can only be read fast enough for roughly 1.5 tokens/sec.
Which runtime should I use?
llama.cpp (CPU build) or Ollama — both run without a GPU
Other RAM Capacities
Other Models on 192 GB system RAM
Devstral on GPUs
What This Model Is Good At
Model & Tools
← Can I Run It? | Devstral model page | Check your hardware