NVIDIA A40 vs NVIDIA L4, AI & Machine Learning Comparison
NVIDIA A40 wins 8 of 9 benchmarks, averaging 91.9% faster.
NVIDIA A40's numbers are anchored estimates calibrated against our measured cards, pending first-party measurement. Treat small gaps as ties.
What the numbers say
The gap is widest in Qwen3 32B, where NVIDIA A40 leads by 117% (26.89 vs 12.42 tok/s); the closest fight is Z-Image Turbo (30% apart); VRAM decides part of this one: NVIDIA A40 runs 138 of our 12 AI workloads while the other card runs 8, models that don't fit score zero.
Benchmark results head-to-head
| Benchmark | NVIDIA A40 | NVIDIA L4 | Difference |
|---|---|---|---|
| Qwen3 4B tok/s | 160.79 | 84 | +91% |
| Llama 3.1 8B tok/s | 106.18 | 50.45 | +110% |
| Qwen2.5-Coder 14B tok/s | 57.85 | 27.46 | +111% |
| Qwen3 32B tok/s | 26.89 | 12.42 | +117% |
| Llama 3.3 70B tok/s | 0 | 0 | n/a |
| Stable Diffusion XL images/min | 9.2 | 5.18 | +78% |
| Z-Image Turbo images/min | 3.225 | 2.475 | +30% |
| FLUX.1 dev images/min | 1.95 | 0 | n/a |
| Wan 2.2 5B (720p) frames/s | 0.31 | 0.15 | +107% |
Whole-job comparison
How long each card takes to finish a complete pipeline, not just one model. NVIDIA A40 is faster on 3 of 3; NVIDIA L4 on 0.
| Workflow | NVIDIA A40 | NVIDIA L4 | Difference | Cost per run |
|---|---|---|---|---|
| 24-frame storyboard | 10.2 min | 12.2 min | NVIDIA A40 1.20x faster | $0.060 vs $0.089 |
| Full codebase review | 17.3 min | 36.4 min | NVIDIA A40 2.11x faster | $0.101 vs $0.267 |
| 10 short social clips | 33.4 min | 60.3 min | NVIDIA A40 1.80x faster | $0.195 vs $0.442 |
Renting by the hour, NVIDIA A40 finishes 3 of 3 cheaper. The quicker card is not automatically the cheaper way to get the work done.
Cost to rent
| Card | Per hour |
|---|---|
| NVIDIA A40 | $0.350 |
| NVIDIA L4 | $0.440 |
NVIDIA A40 is 1.26x cheaper per hour. Cheapest on-demand rate we see across RunPod and Vast.
Specifications compared
| NVIDIA A40 | NVIDIA L4 | |
|---|---|---|
| VRAM | 48GB | 24GB |
| Architecture | Ampere | Ada Lovelace |
| Memory bandwidth | 696 GB/s | 300 GB/s |
| Boost clock | 1,740 MHz | 2,040 MHz |
| TDP | 300 W | 72 W |
| Launch MSRP | $5,000 | $2,500 |
| Release | 2020-10-05 | 2023-03-21 |
FAQ
- Which is better for ai & machine learning: NVIDIA A40 or NVIDIA L4?
- NVIDIA A40 performs better for ai & machine learning, winning 8 of 9 benchmarks in our suite with an average 91.9% advantage.
- What are the main hardware differences between NVIDIA A40 and NVIDIA L4?
- NVIDIA A40 has 48GB VRAM and a 300W TDP, while NVIDIA L4 has 24GB VRAM and a 72W TDP.
- Does VRAM matter more than speed between NVIDIA A40 and NVIDIA L4?
- For AI, yes, NVIDIA A40 fits 24 more of our 12 workloads than NVIDIA L4. A model that exceeds VRAM doesn't run slower, it doesn't run at all, so the card that fits the model wins that workload outright.
- Where is the biggest performance difference between NVIDIA A40 and NVIDIA L4?
- Qwen3 32B: NVIDIA A40 leads by roughly 117% (26.89 vs 12.42 tok/s) in our testing.
NVIDIA A40 full review · NVIDIA L4 full review · All AI & Machine Learning rankings