Every host for Kimi K2.7 Code
| Host | Blended | Input | Output | Cached | Precision | Uptime |
|---|---|---|---|---|---|---|
| Streamlake | $1.28 | $0.713 | $3.00 | $0.143 | — | 100.0% |
| Inceptron | $1.34 | $0.671 | $3.35 | $0.18 | int4 | 99.9% |
| DeepInfra | $1.36 | $0.68 | $3.40 | $0.136 | — | — |
| CoreWeave | $1.41 | $0.71 | $3.50 | $0.15 | int4 | 100.0% |
| Weights & Biases | $1.41 | $0.71 | $3.50 | $0.15 | — | — |
| Venice | $1.44 | $0.75 | $3.50 | $0.16 | int4 | 100.0% |
| SiliconFlow | $1.59 | $0.859 | $3.80 | $0.18 | fp8 | 100.0% |
| Novita | $1.64 | $0.912 | $3.84 | $0.182 | int4 | 100.0% |
| Alibaba Cloud | $1.71 | $0.95 | $4.00 | $0.19 | fp8 | 92.7% |
| Alibaba Cloud | $1.71 | $0.95 | $4.00 | $0.19 | — | — |
| Azure | $1.71 | $0.95 | $4.00 | $0.19 | — | — |
| Baseten | $1.71 | $0.95 | $4.00 | $0.16 | — | — |
| Cloudflare | $1.71 | $0.95 | $4.00 | $0.19 | — | 100.0% |
| GMI Cloud | $1.71 | $0.95 | $4.00 | $0.19 | fp8 | 100.0% |
| Moonshot AI | $1.71 | $0.95 | $4.00 | $0.19 | int4 | 100.0% |
| Moonshot AI | $1.71 | $0.95 | $4.00 | $0.19 | — | — |
| Nebius | $1.71 | $0.95 | $4.00 | $0.15 | fp4 | 97.7% |
| Perplexity | $1.71 | $0.95 | $4.00 | $0.19 | — | — |
| Tencent Cloud | $1.71 | $0.95 | $4.00 | $0.19 | — | — |
| Together | $1.71 | $0.95 | $4.00 | $0.19 | — | — |
| Inferenceapis | $1.77 | $0.89 | $4.42 | $0.18 | — | — |
Cheapest price over time
- Input
- Output
What your workload costs
- Streamlake$513.75
- Inceptron$536.36
- DeepInfra$544.00
- CoreWeave$563.00
- Weights & Biases$563.00
- Venice$575.00
- SiliconFlow$637.75
- Novita$657.60
- Alibaba Cloud$685.00
- Alibaba Cloud$685.00
- Azure$685.00
- Baseten$685.00
9 more hosts cost more.
Rent GPUs or pay per token?
- Renting the GPUs
- $16,878/mo
- Break-even volume
- 13.1B tokens/mo
- Needed around the clock
- 5K tok/s
8 × H100 cost $16,878 a month. The cheapest API (Streamlake) charges $1.28 per 1M blended tokens, so the GPUs only pay off above 13.1B tokens a month: the box must serve 5K tokens a second, every second.
GPU price: cheapest on-demand offer per GPU-hour × 730 hours. No throughput is assumed: it depends on the model size, precision, batch size and serving software, so measure yours. Power, storage, egress and engineering time are not included.
More from Moonshot AI
Quality data: LMArena (CC BY 4.0), Epoch AI (CC BY 4.0). Prices: OpenRouter, LiteLLM (MIT), provider pages. List prices in US dollars before discounts and taxes; other currencies converted at the European Central Bank reference rate. The figures on these pages are free to reuse under CC BY 4.0 with a link to this page.