This GPU offers reliable throughput for local AI workloads. Pair it with the right model quantization to hit your desired tokens/sec, and monitor prices below to catch the best deal.
Quick Answer: NVIDIA L40S has 48GB VRAM, enough for models up to roughly 120B parameters at 4-bit quantization. It draws 350W under load.
With 48GB VRAM, NVIDIA L40S can run models up to approximately 120B parameters using 4-bit quantization. That covers most popular models, including 70B-class ones at 4-bit.
Consider H100 or MI300X — Maximum VRAM for enterprise workloads.
Buy directly on Amazon with fast shipping and reliable customer service.
Essential accessories to pair with NVIDIA L40S
Accessories Total
Typical prices — check Amazon for current pricing
💡 Not ready to buy? Try cloud GPUs first
Test NVIDIA L40S performance in the cloud before investing in hardware. Pay by the hour with no commitment.
Showing 12 of 80 rows. Speeds are calculated estimates, not measurements — search for your model to jump straight to it.
| Model | Size | Quantization | Tokens/sec | VRAM used |
|---|---|---|---|---|
| Deepseek AI Deepseek Coder 1.3B Instruct | 1.3B | Q4 | ~250 tok/sEstimated | 1GB |
| Deepseek AI Deepseek R1 Distill Qwen 1.5B | 1.5B | Q4 | ~250 tok/sEstimated | 1GB |
| Deepseek AI Deepseek Ocr 2 | Unknown | Q4 | ~205 tok/sEstimated | 2GB |
| Deepseek AI Deepseek Ocr | Unknown | Q4 | ~205 tok/sEstimated | 2GB |
| Lmstudio Community Deepseek R1 0528 Qwen3 8B Mlx 8bit | 8B | Q4 | ~205 tok/sEstimated | 4GB |
| Lmstudio Community Deepseek R1 0528 Qwen3 8B Mlx 4bit | 8B | Q4 | ~205 tok/sEstimated | 4GB |
| Deepseek AI Deepseek R1 Distill Qwen 7B | 7B | Q4 | ~205 tok/sEstimated | 4GB |
| Nineninesix Kani Tts 2 En | Unknown | Q4 | ~200 tok/sEstimated | 1GB |
| Qwen Qwen3 Tts 12hz 1 7B Customvoice | 7B | Q4 | ~200 tok/sEstimated | 1GB |
| Zai Org Glm Ocr | Unknown | Q4 | ~200 tok/sEstimated | 1GB |
| Qwen Qwen3 Asr 1 7B | 7B | Q4 | ~200 tok/sEstimated | 2GB |
| Nari Labs Dia2 2B | 2B | Q4 | ~200 tok/sEstimated | 1GB |
Showing 12 of 240 rows.
| Model | Size | Quantization | Verdict | Estimated speed | VRAM needed |
|---|---|---|---|---|---|
| 01 AI Yi 1 5 34B Chat | 34B | Q4 | Fits comfortably | ~58 tok/sEstimated | 18GB (have 48GB) |
| 01 AI Yi 1 5 34B Chat | 34B | Q8 | Fits comfortably | ~41 tok/sEstimated | 35GB (have 48GB) |
| 01 AI Yi 1 5 34B Chat | 34B | FP16 | Not supported | ~22 tok/sEstimated | 69GB (have 48GB) |
| AI Forever Rugpt 3.5 13B | 13B | Q4 | Fits comfortably | ~125 tok/sEstimated | 7GB (have 48GB) |
| AI Forever Rugpt 3.5 13B | 13B | Q8 | Fits comfortably | ~87 tok/sEstimated | 13GB (have 48GB) |
| AI Forever Rugpt 3.5 13B | 13B | FP16 | Fits comfortably | ~47 tok/sEstimated | 26GB (have 48GB) |
| AI Mo Kimina Prover 72B | 72B | Q4 | Fits comfortably | ~33 tok/sEstimated | 37GB (have 48GB) |
| AI Mo Kimina Prover 72B | 72B | Q8 | Not supported | ~23 tok/sEstimated | 73GB (have 48GB) |
| AI Mo Kimina Prover 72B | 72B | FP16 | Not supported | ~13 tok/sEstimated | 146GB (have 48GB) |
| Alibaba Nlp Gte Qwen2 1.5B Instruct | 1.5B | Q4 | Fits comfortably | ~200 tok/sEstimated | 1GB (have 48GB) |
| Alibaba Nlp Gte Qwen2 1.5B Instruct | 1.5B | Q8 | Fits comfortably | ~140 tok/sEstimated | 2GB (have 48GB) |
| Alibaba Nlp Gte Qwen2 1.5B Instruct | 1.5B | FP16 | Fits comfortably | ~76 tok/sEstimated | 4GB (have 48GB) |
Note: Performance estimates are calculated. Real results may vary. Methodology · Submit real data
Explore how RTX 4090 stacks up for local inference workloads.
Explore how RTX 4080 stacks up for local inference workloads.
Explore how RTX 4070 Ti stacks up for local inference workloads.
Explore how RTX 3090 stacks up for local inference workloads.
Explore how RX 7900 XTX stacks up for local inference workloads.