Kimi-K2-Base: GPU memory and hosting cost
Open weights · 1026.41 billion parameters · Mixture of experts · 131,072-token context · licence other
No cloud sells this model per token: to use it you run it yourself, on GPU instances you rent.
Run it yourself
It needs 1283 GB of GPU memory at fp8, the precision it ships in.
| Cloud | Cheapest instance that fits | GPUs | An hour | A month, around the clock |
|---|---|---|---|---|
| Crusoe | mi300x.8x | 8 × MI300X 192 GB | $27.60 | $20,148 |
| OCI | BM.GPU.MI300X.8 | 8 × MI300X 192 GB | $48.00 | $35,040 |
| Together AI | hgx-b200.8gpu | 8 × B200 180 GB | $65.52 | $47,830 |
| Nebius | gpu-b200-sxm-a.8gpu-160vcpu-1792gb | 8 × B200 180 GB | $68.00 | $49,640 |
| Azure | Standard_ND96isr_MI300X_v5 | 8 × MI300X 192 GB | $68.64 | $50,107 |
| AWS | 3 × p4de.24xlarge, joined by EFA | 8 × A100 80 GB in each | $82.34 | $60,109 |
| Alibaba Cloud | 2 × ecs.gn8v-8x.16xlarge, joined by eRDMA | 8 × GPU H 96 GB in each | $116.98 | $85,397 |
| Google Cloud | 2 × a3-ultragpu-8g, joined by GPUDirect RDMA | 8 × H200 141 GB in each | $169.61 | $123,818 |
Several servers: the model is sized across identical servers that hold it together, on instance types whose vendor documents a network between servers; the price is all of them. It is shown where no one server holds the model, or where it costs less than the one that does.
More from Moonshot AI
Similar size, open weights
List prices collected 4 November 2026, refreshed every three hours. No negotiated rates or taxes.