A40 48G — what LLMs can it run?
94 of 110 indexed models fit comfortably in 48GB at 4K context, each at the highest-quality quant that still leaves headroom.
The short answer for A40 48G
48 GB, 696 GB/s GDDR6. 94 of 110 models in this index fit comfortably at 4K context; 95 load at all.
- Biggest that fits
- GLM-4.5-Air — 37.2 GB at Q2_K — 10.8 GB spare at 4K, so longer context comes out of a thin margin.
- Room to grow
- Nemotron 3.5 Bolt 74B-A9B — 23.1 GB at IQ2_XXS — under 60% of the card, which leaves 25.0 GB for a long context window or a second process.
- Speed ceiling
- Llama 3.1 8B Instruct at Q4_K_M reads 4.6 GB of weights per token, so 696 GB/s puts a hard ceiling near 151 tok/s. That is arithmetic on two published numbers, not a benchmark — real throughput lands below it. No run on this card has been measured here, so there is nothing to compare the ceiling against.
- The wall you will hit
- Capacity. 94 of 110 models clear 48 GB comfortably at 4K context, and the ones that do not are short by whole gigabytes rather than by a rounding error — at 696 GB/s the memory system is not what holds this card back. Moving up a size class is what changes the list, not a different quant.
70B+ · 7
| Model | Quant | Est. VRAM | Headroom | tok/s |
|---|---|---|---|---|
| GLM-4.5-Air106B MoE | Q2_K | 37.2 GB | +10.8 GB | ≤186Ceiling |
| Nemotron 3.5 Bolt 74B-A9B74B-A9B | IQ2_XXS | 23.05 GB | +25 GB | ≤274Ceiling |
| Qwen2.5 72B Instruct72B | AWQ INT4 | 39.36 GB | +8.6 GB | ≤20Ceiling |
| Llama 3.1 70B Instruct70B | AWQ INT4 | 38.26 GB | +9.7 GB | ≤21Ceiling |
| Llama 3.3 70B Instruct70B | AWQ INT4 | 38.26 GB | +9.7 GB | ≤21Ceiling |
| DeepSeek-R1-Distill-Llama-70B70B | AWQ INT4 | 38.26 GB | +9.7 GB | ≤21Ceiling |
| Jamba 1.5 Mini52B-A12B | Q4_K_M | 33.01 GB | +15 GB | ≤101Ceiling |
32B · 30
| Model | Quant | Est. VRAM | Headroom | tok/s |
|---|---|---|---|---|
| Kimi Linear 48B-A3B Instruct48B-A3B | Q4_K_M | 30.44 GB | +17.6 GB | ≤403Ceiling |
| Mixtral 8x7B Instruct47B MoE | Q4_K_M | 30.13 GB | +17.9 GB | ≤94Ceiling |
| Seed-OSS 36B Instruct36B | Q5_K_M | 27.81 GB | +20.2 GB | ≤29Ceiling |
| Qwen3.6 35B-A3B35B-A3B | Q8_0 | 39.6 GB | +8.4 GB | ≤230Ceiling |
| Command R 35B35B | Q4_K_M | 22.86 GB | +25.1 GB | ≤35Ceiling |
| Yi 1.5 34B Chat34B | Q4_K_M | 22.82 GB | +25.2 GB | ≤35Ceiling |
| Qwen3 32B Instruct32B | Q4_K_M | 21.85 GB | +26.2 GB | 24–31Derived |
| Qwen2.5 32B Instruct32B | Q4_K_M | 21.69 GB | +26.3 GB | 24–32Derived |
| Qwen2.5-Coder 32B Instruct32B | Q4_K_M | 21.69 GB | +26.3 GB | 24–32Derived |
| DeepSeek-R1-Distill-Qwen-32B32B | Q4_K_M | 21.69 GB | +26.3 GB | 24–32Derived |
| Nemotron 3.5 Lightning 30B-A3B30B-A3B | Q8_0 | 35.14 GB | +12.9 GB | ≤230Ceiling |
| Gemma 4 31B IT31B | Q8_0 | 35.72 GB | +12.3 GB | 14–19Derived |
| Qwen3 30B-A3B Instruct30B-A3B | Q5_K_M | 23.04 GB | +25 GB | ≤344Ceiling |
| Qwen3-Coder 30B-A3B Instruct30B-A3B | Q5_K_M | 23.04 GB | +25 GB | ≤344Ceiling |
| Qwen3-VL 30B-A3B Instruct30B-A3B | Q8_0 | 34.28 GB | +13.7 GB | ≤230Ceiling |
| GLM-4.7-Flash30B-A3B | Q8_0 | 33.47 GB | +14.5 GB | ≤230Ceiling |
| Granite 4.2 30B30B | Q8_0 | 33.61 GB | +14.4 GB | 15–20Derived |
| Muse Glimmer 30B30B | Q8_0 | 31.08 GB | +16.9 GB | 16–21Derived |
| Qwen3.8 27B27B | Q4_K_M | 17.47 GB | +30.5 GB | 29–38Derived |
| Qwen3.6 27B27B | Q8_0 | 30.71 GB | +17.3 GB | 16–21Derived |
| Gemma 3 27B IT27B | Q4_K_M | 18.37 GB | +29.6 GB | 28–37Derived |
| Gemma 2 27B Instruct27B | Q5_K_M | 21.76 GB | +26.2 GB | 24–32Derived |
| Gemma 4 26B-A4B IT26B-A4B | Q8_0 | 28.42 GB | +19.6 GB | ≤172Ceiling |
| Mistral Small 24B Instruct24B | EXL2 4.65bpw | 15.26 GB | +32.7 GB | ≤53Ceiling |
| LFM2.5 24B-A2B24B-A2B | Q8_0 | 26.73 GB | +21.3 GB | ≤281Ceiling |
| Devstral Small 1.1 24B24B | Q6_K | 20.91 GB | +27.1 GB | 24–32Derived |
| Magistral Small 1.2 24B24B | Q6_K | 20.91 GB | +27.1 GB | 24–32Derived |
| Codestral 22B22B | Q4_K_M | 15.03 GB | +33 GB | 35–46Derived |
| ERNIE 4.5 21B-A3B21B-A3B | Q8_0 | 24.44 GB | +23.6 GB | ≤230Ceiling |
| GPT-OSS 20B21B MoE | MXFP4 | 12.76 GB | +35.2 GB | ≤352Ceiling |
14B · 18
| Model | Quant | Est. VRAM | Headroom | tok/s |
|---|---|---|---|---|
| InternLM2 20B Chat20B | Q5_K_M | 15.59 GB | +32.4 GB | 33–44Derived |
| DeepSeek-Coder-V2-Lite Instruct16B-A2.4B | Q8_0 | 17.56 GB | +30.4 GB | ≤287Ceiling |
| DeepSeek-V2-Lite Chat16B-A2.4B | Q4_K_M | 10.08 GB | +37.9 GB | ≤504Ceiling |
| StarCoder2 15B15B | Q4_K_M | 10.19 GB | +37.8 GB | 52–68Derived |
| Qwen3 14B Instruct14B | Q5_K_M | 11.65 GB | +36.4 GB | 45–59Derived |
| Qwen2.5 14B Instruct14B | Q5_K_M | 11.73 GB | +36.3 GB | 45–60Derived |
| DeepSeek-R1-Distill-Qwen-14B14B | EXL2 4.65bpw | 9.75 GB | +38.3 GB | ≤86Ceiling |
| Phi-4 14B14B | Q5_K_M | 11.77 GB | +36.2 GB | 45–60Derived |
| Phi-3 Medium 14B Instruct14B | Q6_K | 12.86 GB | +35.1 GB | 41–54Derived |
| Mistral Nemo 12B Instruct12B | Q6_K | 11.14 GB | +36.9 GB | 47–62Derived |
| Gemma 3 12B IT12B | Q5_K_M | 9.84 GB | +38.2 GB | 54–72Derived |
| Mellum2.1 12B-A2.5B Thinking12B-A2.5B | Q8_0 | 13.6 GB | +34.4 GB | ≤276Ceiling |
| Stable LM 2 12B Chat12B | Q4_K_M | 8.35 GB | +39.7 GB | 64–85Derived |
| Llama 3.2 11B Vision Instruct11B | Q8_0 | 12.9 GB | +35.1 GB | 40–53Derived |
| Solar 10.7B Instruct11B | Q4_K_M | 7.6 GB | +40.4 GB | 72–96Derived |
| Falcon 3 10B Instruct10B | Q4_K_M | 7.28 GB | +40.7 GB | 75–99Derived |
| Gemma 2 9B Instruct9B | Q8_0 | 11.7 GB | +36.3 GB | 48–63Derived |
| Qwen3.5 9B9B | Q8_0 | 10.37 GB | +37.6 GB | 48–63Derived |
7B · 25
| Model | Quant | Est. VRAM | Headroom | tok/s |
|---|---|---|---|---|
| GLM-4-9B-Chat9B | Q8_0 | 10.16 GB | +37.8 GB | 49–65Derived |
| Qwen3-VL 8B Instruct8B | Q8_0 | 10.39 GB | +37.6 GB | 50–67Derived |
| Granite 4.2 8B8B | Q8_0 | 10.45 GB | +37.6 GB | 50–67Derived |
| Ministral 3 8B Instruct8B | Q8_0 | 10.02 GB | +38 GB | 52–69Derived |
| Qwen2-VL 7B Instruct7B | Q4_K_M | 5.49 GB | +42.5 GB | 94–124Derived |
| Granite 3.1 8B Instruct8B | Q4_K_M | 5.88 GB | +42.1 GB | 95–125Derived |
| Qwen3 8B Instruct8B | Q6_K | 7.64 GB | +40.4 GB | 70–93Derived |
| Llama 3.1 8B Instruct8B | Q8_0 | 9.47 GB | +38.5 GB | 55–73Derived |
| Nous Hermes 3 Llama 3.1 8B8B | EXL2 4.65bpw | 5.43 GB | +42.6 GB | ≤157Ceiling |
| Aya 23 8B8B | Q4_K_M | 5.64 GB | +42.4 GB | 97–128Derived |
| OpenChat 3.6 8B8B | EXL2 4.65bpw | 5.43 GB | +42.6 GB | ≤157Ceiling |
| DeepSeek-R1-Distill-Llama-8B8B | Q5_K_M | 6.51 GB | +41.5 GB | 82–109Derived |
| InternLM2 7B Chat7B | Q4_K_M | 5.45 GB | +42.6 GB | 100–132Derived |
| Qwen2.5 7B Instruct7B | Q6_K | 6.77 GB | +41.2 GB | 75–99Derived |
| Qwen2.5-Coder 7B Instruct7B | EXL2 4.65bpw | 4.87 GB | +43.1 GB | ≤165Ceiling |
| DeepSeek-R1-Distill-Qwen-7B7B | EXL2 4.65bpw | 4.87 GB | +43.1 GB | ≤165Ceiling |
| Gemma 4 E4B ITE4B | Q8_0 | 8.46 GB | +39.5 GB | 59–78Derived |
| OLMo 2 7B Instruct7B | Q8_0 | 10.3 GB | +37.7 GB | 61–80Derived |
| OLMo 3 7B Instruct7B | Q8_0 | 10.3 GB | +37.7 GB | 61–80Derived |
| WizardLM-2 7B7B | Q4_K_M | 5.14 GB | +42.9 GB | 107–142Derived |
| Mistral 7B Instruct v0.37B | Q6_K | 6.75 GB | +41.3 GB | 79–105Derived |
| Zephyr 7B Beta7B | Q6_K | 6.75 GB | +41.3 GB | 79–105Derived |
| Gemma 4 E2B ITE2B | Q8_0 | 5.2 GB | +42.8 GB | ≤148Ceiling |
| Qwen3.5 4B4B | Q8_0 | 4.97 GB | +43 GB | ≤159Ceiling |
| Gemma 3 4B IT4B | Q8_0 | 5.05 GB | +43 GB | ≤160Ceiling |
≤3B · 14
| Model | Quant | Est. VRAM | Headroom | tok/s |
|---|---|---|---|---|
| Qwen3 4B Instruct4B | Q6_K | 4.05 GB | +44 GB | ≤223Ceiling |
| Phi-4 Mini Instruct3.8B | Q8_0 | 4.81 GB | +43.2 GB | ≤179Ceiling |
| Phi-3.5 Mini Instruct3.8B | Q8_0 | 5.89 GB | +42.1 GB | ≤180Ceiling |
| Granite 4.2 3B3B | Q8_0 | 4.41 GB | +43.6 GB | ≤189Ceiling |
| Llama 3.2 3B Instruct3B | Q8_0 | 4.05 GB | +44 GB | ≤215Ceiling |
| Qwen2.5 3B Instruct3B | Q8_0 | 3.59 GB | +44.4 GB | ≤223Ceiling |
| SmolLM3 3B3B | Q8_0 | 3.73 GB | +44.3 GB | ≤224Ceiling |
| Gemma 2 2B Instruct2B | Q8_0 | 3.34 GB | +44.7 GB | ≤265Ceiling |
| Qwen3.5 2B2B | Q8_0 | 2.23 GB | +45.8 GB | ≤352Ceiling |
| Qwen3 1.7B Instruct1.7B | Q8_0 | 2.39 GB | +45.6 GB | ≤400Ceiling |
| Qwen2.5 1.5B Instruct1.5B | Q8_0 | 1.83 GB | +46.2 GB | ≤449Ceiling |
| Llama 3.2 1B Instruct1B | Q8_0 | 1.51 GB | +46.5 GB | ≤557Ceiling |
| Qwen3.5 0.8B0.8B | Q8_0 | 0.94 GB | +47.1 GB | ≤859Ceiling |
| Qwen2.5 0.5B Instruct0.5B | Q8_0 | 0.6 GB | +47.4 GB | ≤1420Ceiling |
How this list is built
Each row is the lowest-perplexity-loss quant of that model whose estimated total — weights plus KV cache at 4K context plus activation buffer — uses at most 88% of the card. That is the calculator's "green" threshold, so every row here has real headroom rather than only just fitting. Raise the context length and the list shortens; the calculator lets you check any combination directly. The speed column is this card: a measured run, a range scaled from this card's own Llama 2 7B benchmark, or a bandwidth ceiling. A figure that only exists for the RTX 4090 is labelled as that card.
A further 1 models load but with no headroom to spare (up to 105% of VRAM) — the Quant Hub’s GPU chips count those too, which is why its number is higher.
Measured on this card
No benchmark runs in this index were recorded on a A40 48G. Every figure on this page is calculated from the model architecture and the quant level — treat them as estimates, not measurements.
Cards with the same budget
What fits is decided by memory, so every 48GB card of this type returns the same list. These pages are not different answers — they differ in throughput, which this index does not measure per card.
Stepping up
A A100 80G (80GB) fits 4 more of the indexed models than this card. A100 80G →
Before buying, hear the speed difference on a model both cards hold: A40 48G vs A100 80G (Qwen3 32B Instruct) →
Common questions
What is the best local LLM for an A40 48G?
For everyday use, Nemotron 3.5 Bolt 74B-A9B at IQ2_XXS — about 23.1 GB of the card's 48 GB at 4K context, leaving 25.0 GB for a longer window. If you want the largest thing that will load, that is GLM-4.5-Air at Q2_K (37.2 GB, 10.8 GB spare). "Best" here means best fit for the memory budget — this index does not run task benchmarks, so it cannot tell you which model is smarter.
Can an A40 48G run Llama 4 Scout 17B (16E)?
No. Its smallest build here, Q3_K_M, needs about 55.9 GB and this card has 48.0 GB usable — short by 7.9 GB before any context beyond 4K. The largest model this card does clear is GLM-4.5-Air.
How many tokens per second does an A40 48G do on an 8B model at Q4?
This index has no measured run on this card, so it does not publish a figure. What can be stated from specifications: generating a token requires reading every weight once, Llama 3.1 8B Instruct at Q4_K_M is 4.6 GB of weights, and this card moves 696 GB/s — a ceiling near 151 tok/s. Batch size, context length, the runtime and how much of the model sits in cache all take you below it.
A40 48G or L40S 48G for local LLMs?
They hold the same models: 48 GB against 48 GB fits 94 and 94 of 110 respectively at 4K. The difference is throughput — 696 GB/s against 864 GB/s, a 1.2× gap in how fast the weights can be read, which is what token generation is bound by. The L40S 48G generates faster on any model both can hold.
94 of 110 indexed models fit comfortably in 48GB at 4K context, each at the highest-quality quant that still leaves headroom.
This page's figures change when the model or the runtime does.Last updated 2026-10-18 RSS → /feed.xml