Publicly listed pricing monitoring results across platforms on October 18 (auto-refreshed hourly; full dataset available atReal-time price comparison homepage; prices are subject to the order page.
| GPU model | Today’s lowest monitored price | 24 hours | Platform | vs. yesterday | |
|---|---|---|---|---|---|
| RTX 3090 | ¥0.63/hour | ¥15/day | Vast.ai (USA) | ▼59.9% | Rent Now |
| V100 16GB | ¥0.80/hour | ¥19/day | Verda (Finland) | No change | Rent Now |
| RTX 4090 | ¥0.87/hour | ¥21/day | Vast.ai (USA) | No change | Rent Now |
| RTX 4090D 24GB | ¥1.14/hour | ¥27/day | Suanjiayun | No change | Rent Now |
| T4 16GB | ¥1.35/hour | ¥32/day | MatPool | No change | Rent Now |
| RTX 5090 | ¥1.53/hour | ¥37/day | Vast.ai (USA) | No change | Rent Now |
| A6000 48GB | ¥2.20/hour | ¥53/day | Gpushare | No change | Rent Now |
| A40 48GB | ¥2.52/hour | ¥60/day | RunPod (USA) | No change | Rent Now |
| L4 24GB | ¥3.17/hour | ¥76/day | RunPod (USA) | No change | Rent Now |
| H20 96GB | ¥3.40/hour | ¥82/day | PPIO Paiou Cloud | No change | Rent Now |
| MI300X 192GB | ¥3.60/hour | ¥86/day | RunPod (USA) | No change | Rent Now |
| RTX 6000 Ada 48GB | ¥4.78/hour | ¥115/day | Verda (Finland) | No change | Rent Now |
| L40 48GB | ¥4.97/hour | ¥119/day | RunPod (USA) | No change | Rent Now |
| A800 80GB | ¥5.15/hour | ¥124/day | MornAI | No change | Rent Now |
| L40S 48GB | ¥5.69/hour | ¥137/day | RunPod (USA) | No change | Rent Now |
| A100 80GB | ¥7.33/hour | ¥176/day | Verda (Finland) | No change | Rent Now |
| RTX PRO 6000 96GB | ¥8.45/hour | ¥203/day | Verda (Finland) | No change | Rent Now |
| H100 80GB | ¥10.37/hour | ¥249/day | Vast.ai (USA) | ▼27.1% | Rent Now |
| H200 141GB | ¥19.97/hour | ¥479/day | Verda (Finland) | No change | Rent Now |
| B200 180GB | ¥28.76/hour | ¥690/day | Verda (Finland) | No change | Rent Now |
Today’s Insights
· RTX 3090: only ¥15/day (Vast.ai, US); down 59.9%; top choice for fine-tuning 7B/13B models; ample stock available.
· H100: ¥249/day (Vast.ai, US); down 27.1%; significant reduction in LLM inference/training costs; ideal for short-term high-load tasks.
· A100 80GB: ¥176/day (Verda, Finland); price stable; remains the cost-efficiency benchmark for multi-GPU cluster training.
· RTX 4090: ¥21/day (Vast.ai, US); notable price gap vs. RTX 4090D (¥27/day at Suanjiayun); prefer US-based spot instances for lightweight training.
· L40S/RTX 6000 Ada and other high-end GPUs: over ¥115/day; recommended only for FP8 LLM inference or memory-sensitive workloads.
· T4 (¥32/day at MatPool) and V100 (¥19/day at Verda, Finland) price gap narrows; consider V100 for legacy model compatibility.