Moonshot AI models: prices and the hardware each needs
11 models, 4 sold per token by 12 vendors. Vendors are in alphabetical order; the price is the cheapest list price among them, per million tokens.
| Model | Size | Sold per token by | From, input | Output | GPU memory |
|---|---|---|---|---|---|
| Kimi K2.5 | 1027 B | Alibaba Cloud, AWS, Azure, Cloudflare, Core42 | $0.574 | $3.01 | 744 GB |
| Kimi K2.6 | 1027 B | Azure, Cloudflare, DeepInfra, Nebius, Telnyx | $0.75 | $3.50 | 744 GB |
| Kimi K2.7 Code | 1027 B | Alibaba Cloud, Azure, Cloudflare, Nebius | $0.894 | $3.71 | 744 GB |
| Kimi K3 | 2780 B | Alibaba Cloud, AWS, Azure, Databricks, DeepInfra, Fireworks AI, Nebius, Snowflake, Telnyx, Together AI | $2.70 | $13.50 | 1951.1 GB |
| Kimi-K2-Base | 1026.41 B | not sold per token | 1283 GB | ||
| Kimi-K2-Instruct | 1026.41 B | not sold per token | 1283 GB | ||
| Kimi-K2-Instruct-0905 | 1026.41 B | not sold per token | 1283 GB | ||
| Kimi-Linear-48B-A3B-Base | 49.12 B | not sold per token | 122.8 GB | ||
| Kimi-Linear-48B-A3B-Instruct | 49.12 B | not sold per token | 122.8 GB | ||
| Moonlight-16B-A3B | 15.96 B | not sold per token | 39.9 GB | ||
| Moonlight-16B-A3B-Instruct | 15.96 B | not sold per token | 39.9 GB |