Llama on NVIDIA Ampere
Can I run Llama 4 Scout 109B-A17B on an RTX A6000?
Not at Q4_K_M — it needs 64.0 GB against 44.3 GB available. Drop to IQ3_XXS and it fits, at about 58.2 tokens per second.
Won't fit
64.0 GB of 44.3 GB · 144%067 GB
Weights
61.7 GB
KV cache
1.5 GB
Runtime overhead
0.8 GB
Over the limit
19.7 GB
Short by 19.7 GB. You can run it with 32 of 48 layers on the RTX A6000 and the rest in system RAM, at roughly 9.64 tok/s — usable for batch work, painful for chat. A smaller quantisation or a shorter context is usually the better trade.
Generation9.64tok/s
Prompt processing926tok/s
Max context0tokens
KV per 1K tokens0GB
Every quantisation of Llama 4 Scout 109B-A17B on a RTX A6000
Highlighted row is the highest quality that still fits at 8K context.
| Quantisation | bpw | Weights | Total | Verdict | Max ctx | tok/s |
|---|---|---|---|---|---|---|
| FP16 / BF16 | 16.00 | 203.0 GB | 205.4 GB | Won't fit | — | 1.48 |
| INT8 / W8A8 | 8.50 | 107.4 GB | 109.7 GB | Won't fit | — | 3.45 |
| Q8_0 (GGUF) | 8.50 | 107.4 GB | 109.7 GB | Won't fit | — | 3.45 |
| FP8 (E4M3) | 8.00 | 101.5 GB | 103.9 GB | Won't fit | — | 3.74 |
| Q6_K | 6.56 | 83.2 GB | 85.6 GB | Won't fit | — | 5.29 |
| Q5_K_M | 5.67 | 71.9 GB | 74.3 GB | Won't fit | — | 7.03 |
| Q5_K_S | 5.52 | 70.0 GB | 72.4 GB | Won't fit | — | 7.21 |
| Q4_K_M | 4.85 | 61.7 GB | 64.0 GB | Won't fit | — | 9.64 |
| Q4_K_S | 4.58 | 58.3 GB | 60.7 GB | Won't fit | — | 11.2 |
| Q4_0 | 4.55 | 58.0 GB | 60.3 GB | Won't fit | — | 11.3 |
| AWQ 4-bit | 4.25 | 56.8 GB | 59.1 GB | Won't fit | — | 12.1 |
| GPTQ 4-bit | 4.25 | 56.8 GB | 59.1 GB | Won't fit | — | 12.1 |
| MXFP4 | 4.25 | 56.8 GB | 59.1 GB | Won't fit | — | 12.1 |
| IQ4_XS | 4.25 | 54.2 GB | 56.6 GB | Won't fit | — | 14.2 |
| Q3_K_M | 3.91 | 50.0 GB | 52.3 GB | Won't fit | — | 18.7 |
| IQ3_M | 3.70 | 47.4 GB | 49.7 GB | Won't fit | — | 23.1 |
| IQ3_XXS | 3.06 | 39.4 GB | 41.8 GB | Fits, but tight | 22K | 58.2 |
| Q2_K | 2.63 | 34.1 GB | 36.4 GB | Runs comfortably | 50K | 65.9 |
| IQ2_XXS | 2.06 | 27.0 GB | 29.3 GB | Runs comfortably | 88K | 80.0 |
| IQ1_M | 1.75 | 23.1 GB | 25.4 GB | Runs comfortably | 109K | 90.6 |