作者： Jakub Rusinowski · 最后更新： 2026年4月5日
Yes, comfortably — you'll have ~19 GB of headroom running Devstral-2 22B at Q4_K_M (13 GB, ~138 tok/s (est.)) with room for up to 8K context.
在亚马逊查看价格 — NVIDIA GeForce RTX 5090 32GB
| VRAM | 32 GB |
| Memory Bandwidth | 1792 GB/s |
| Quant | VRAM needed | Fits 32 GB? | Max context |
|---|---|---|---|
| F16 | 51.8 GB | ✗ No | — |
| Q8_0 | 31.2 GB | ✓ Yes | 4K |
| Q6_K | 25.9 GB | ✓ Yes | 4K |
| Q5_K_M | 23.4 GB | ✓ Yes | 8K |
| Q4_K_M | 21.1 GB | ✓ Yes | 8K |
| Q3_K_M | 17.2 GB | ✓ Yes | 8K |
| Q2_K | 15.1 GB | ✓ Yes | 8K |
VRAM needed assumes a 4K-token context with an f16 KV cache; “Max context” is the largest window that still fits in 32 GB. Figures are estimates from parameter count, quantization and memory bandwidth — the analyzer lets you tune KV-cache quant and context.
| Devstral-2 22B | Q4_K_M · 13 GB · ~138 tok/s (est.) |
At 2 hrs/day, buying (~$1,999) beats renting at $0.34/hr after about 8.2 years.
Affiliate links — we may earn a commission if you sign up, at no extra cost to you.
Cloud rates verified 2026-07 — estimates, and marketplace prices vary. Buying price is GPU MSRP only, not a full PC.
Yes, comfortably — you'll have ~19 GB of headroom running Devstral-2 22B at Q4_K_M (13 GB, ~138 tok/s (est.)) with room for up to 8K context.
Devstral-2 22B at Q4_K_M quantization (13 GB), estimated ~138 tokens/sec, up to 8K context.
← Can I Run It? | Devstral-2 Model Page | NVIDIA GeForce RTX 5090 GPU Page | Check Your Hardware