Can I Run Qwen 2.5 Family on NVIDIA L40S?
Written by Jakub Rusinowski · Last updated July 21, 2026
Yes, but it's tight — Qwen 2.5 72B Instruct at Q4_K_M needs 47 GB at 8K context (43.5 GB weights + 3.5 GB KV cache/overhead), leaving ~1 GB of the NVIDIA L40S's 48 GB, at ~17 tok/s (est.), with room for up to 8K context.
See what else this hardware can run →
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.
NVIDIA L40S Specs
| VRAM | 48 GB |
| Memory Bandwidth | 864 GB/s |
Qwen 2.5 72B Instruct on the NVIDIA L40S: VRAM by quantization
| Quant | VRAM needed | Fits 48 GB? | Max context | Whole PC |
|---|---|---|---|---|
| F16 | 147.5 GB | ✗ No | — | — |
| Q8_0 | 80 GB | ✗ No | — | — |
| Q6_K | 62.5 GB | ✗ No | — | — |
| Q5_K_M | 54.5 GB | ✗ No | — | — |
| Q4_K_M | 47 GB | ✓ Yes | 8K | — |
| Q3_K_M | 34.2 GB | ✓ Yes | 32K | — |
| Q2_K | 27.2 GB | ✓ Yes | 64K | — |
Assumes an 8K-token context with an f16 KV cache. A longer window needs more; a quantized KV cache needs less. “Max context” is the largest window that still fits in 48 GB. Figures are estimates from parameter count, quantization and memory bandwidth — the analyzer lets you tune KV-cache quant and context.
Context costs VRAM too. At Q4_K_M on the NVIDIA L40S: 8K 47 GB · 32K 55 GB ✗ · 128K 87.2 GB ✗. Past 32K it no longer fits 48 GB — a q8_0 KV cache buys roughly half of that back, which the calculator will price for you.
No compatible GPU? Qwen 2.5 Family on 8 GB of system RAM, CPU only: runs, but it is tight.
Qwen 2.5 Family Sizes That Fit the NVIDIA L40S (at 8K context)
| Qwen 2.5 72B Instruct | Q4_K_M · 47 GB at 8K context · ~17 tok/s (est.) |
| Qwen 2.5 Coder 32B | Q4_K_M · 22.3 GB at 8K context · ~36 tok/s (est.) |
| Qwen 2.5 14B Instruct | Q4_K_M · 10.9 GB at 8K context · ~72 tok/s (est.) |
| Qwen 2.5 7B Instruct | Q4_K_M · 5.9 GB at 8K context · ~121 tok/s (est.) |
At 2 hrs/day, buying (~$7,499) beats renting at $0.77/hr after about 13.5 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.
FAQ
Is the NVIDIA L40S enough to run Qwen 2.5 Family locally?
Yes, but it's tight — Qwen 2.5 72B Instruct at Q4_K_M needs 47 GB at 8K context (43.5 GB weights + 3.5 GB KV cache/overhead), leaving ~1 GB of the NVIDIA L40S's 48 GB, at ~17 tok/s (est.), with room for up to 8K context.
What's the best Qwen 2.5 Family size for the NVIDIA L40S?
Qwen 2.5 72B Instruct at Q4_K_M quantization needs 47 GB at 8K context (43.5 GB weights + 3.5 GB KV cache/overhead), estimated ~17 tokens/sec, up to 8K context.
Every Qwen 2.5 Family size on the NVIDIA L40S
| Size | VRAM at 8K | Verdict | Speed |
|---|---|---|---|
| Qwen 2.5 72B Instruct | 47 GB | Runs | ~17 tok/s |
| Qwen 2.5 Coder 32B | 22.3 GB | Runs | ~36 tok/s |
| Qwen 2.5 14B Instruct | 10.9 GB | Runs | ~72 tok/s |
| Qwen 2.5 7B Instruct | 5.9 GB | Runs | ~121 tok/s |
Qwen 2.5 Family on Other GPUs
- Qwen 2.5 Family on NVIDIA RTX 6000 Ada Generation
- Qwen 2.5 Family on Apple M3 Pro
- Qwen 2.5 Family on NVIDIA GeForce RTX 5090
- Qwen 2.5 Family on Apple M4
- Qwen 2.5 Family on Apple M2 Pro
Popular Models on the NVIDIA L40S
VRAM Tier
Troubleshooting
- CUDA out of memory — why it happens and how to fix it
- Which GGUF quant should I download? (Q4 vs Q5 vs Q8)
Buying Guide
← Can I Run It? | Qwen 2.5 Family Model Page | NVIDIA L40S GPU Page | VRAM calculator | Check Your Hardware