

NVIDIA A40 wins 7 of 9 benchmarks, averaging 28.7% faster.
NVIDIA A40's numbers are anchored estimates calibrated against our measured cards, pending first-party measurement. Treat small gaps as ties.
The gap is widest in Wan 2.2 5B (720p), where NVIDIA A40 leads by 72% (0.18 vs 0.31 frames/s); the closest fight is Z-Image Turbo (7% apart); VRAM decides part of this one: NVIDIA A40 runs 137 of our 12 AI workloads while the other card runs 8, models that don't fit score zero.
| Benchmark | NVIDIA A10G | NVIDIA A40 | Difference |
|---|---|---|---|
| Qwen3-4B tok/s | 129.87 | 160.79 | -19% |
| Llama-3.1-8B tok/s | 86.62 | 106.18 | -18% |
| Qwen2.5-Coder-14B tok/s | 47.07 | 57.85 | -19% |
| Qwen3-32B tok/s | 22.04 | 26.89 | -18% |
| Llama 3.3 70B tok/s | 0 | 0 | n/a |
| Stable Diffusion XL images/min | 6.38 | 9.2 | -31% |
| Z-Image Turbo images/min | 3.45 | 3.225 | +7% |
| FLUX.1 dev images/min | 0 | 1.95 | n/a |
| Wan 2.2 5B (720p) frames/s | 0.18 | 0.31 | -42% |
How long each card takes to finish a complete pipeline, not just one model. NVIDIA A10G is faster on 1 of 3; NVIDIA A40 on 2.
| Workflow | NVIDIA A10G | NVIDIA A40 | Difference |
|---|---|---|---|
| 24-frame storyboard | 9 min | 10.2 min | NVIDIA A10G 1.14x faster |
| Full codebase review | 21.3 min | 17.3 min | NVIDIA A40 1.23x faster |
| 10 short social clips | 50.7 min | 33.4 min | NVIDIA A40 1.52x faster |
| NVIDIA A10G | NVIDIA A40 | |
|---|---|---|
| VRAM | 24GB | 48GB |
| Architecture | Ampere | Ampere |
| Memory bandwidth | 600 GB/s | 696 GB/s |
| Boost clock | 1,710 MHz | 1,740 MHz |
| TDP | 150 W | 300 W |
| Launch MSRP | $2,800 | $5,000 |
| Release | 2021-11-01 | 2020-10-05 |
NVIDIA A10G full review · NVIDIA A40 full review · All AI & Machine Learning rankings