▸ DEVICE UNDER TEST
NVIDIA A10M — 20 GB VRAM.
▸ A10M SPEC
- BRAND
- NVIDIA
- VRAM
- 20 GB GDDR6
- BANDWIDTH
- 500 GB/s
- FP16 COMPUTE
- 23.4 TFLOPS
- FP32 COMPUTE
- 23.4 TFLOPS
- CUDA CORES
- 7,168
- TENSOR CORES
- 224
- TDP
- 150 W
- ARCHITECTURE
- Ampere
▸ AI CAPABILITY
212/ 331 models @ Q4
With 20 GB VRAM and 500 GB/s bandwidth, this GPU handles models up to 28B parameters.
Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~57 tok/s.
§ 01TOP MODELS FOR A10M
212 FIT · SHOWING 20| MODEL | SIZE | VRAM Q4 | TOK/S | AVG |
|---|---|---|---|---|
| PaliGemma 2 28B | 28B | 17.6 GB | 14 | 38.6 |
| Qwen3.5-27B | 27.8B | 17.5 GB | 14 | 59.4 |
| gemma-3-27b | 27.4B | 17.2 GB | 15 | 27.2 |
| gemma-2-27b | 27.2B | 17.1 GB | 15 | 34.6 |
| TranslateGemma 27B | 27B | 17.0 GB | 15 | 38.6 |
| Gemma 4 26B A4B | 26B | 16.4 GB | 100 | 47.9 |
| Mistral-Small-24B | 24B | 15.2 GB | 17 | 25.0 |
| Mistral-Small-3.1-24B | 24B | 15.2 GB | 17 | 28.8 |
| Magistral Small 24B | 24B | 15.2 GB | 17 | 47.0 |
| Devstral Small 2 24B | 24B | 15.2 GB | 17 | 33.4 |
| Codestral 22B | 22.2B | 14.1 GB | 18 | 50.1 |
| Devstral Small 22B | 22.2B | 14.1 GB | 18 | 35.5 |
| Mistral Small 22B | 22.2B | 14.1 GB | 18 | 35.2 |
| SOLAR-Pro 22B | 22.1B | 14.0 GB | 18 | 44.2 |
| ERNIE 4.5 21B A3B | 21B | 13.3 GB | 133 | — |
| GPT-OSS 20B | 21B | 13.3 GB | 111 | 52.9 |
| InternLM2 20B | 19.8B | 12.6 GB | 20 | 45.1 |
| InternLM2.5 20B | 19.8B | 12.6 GB | 20 | 50.9 |
| Ling-lite 16.8B | 16.8B | 10.8 GB | 167 | — |
| DeepSeek V2 Lite 16B | 16B | 10.3 GB | 167 | 38.0 |