A full local Kimi K2 build with 1T total and 32B active parameters. Strong at long-horizon agents and coding, requiring at least three H200s.
Which GPU does Kimi K2 1T-A32B need?
The entry configuration is 3x H200, which delivers around 19 tokens per second. For more parallel requests and headroom on speed, go with 6x H200.
How much does Kimi K2 1T-A32B cost?
An hour of Kimi K2 1T-A32B runs from $20.28 to $54.78 depending on the GPU configuration. Billing is hourly: you pay for server time, not for tokens.
How do I call Kimi K2 1T-A32B from my code?
Once deployed you get an OpenAI-compatible endpoint: point any OpenAI client at the base_url you receive and pass huihui_ai/kimi-k2:1026b as the model name. Code written against the OpenAI API needs no changes.
Does the data stay private?
Yes. Kimi K2 1T-A32B runs on a dedicated GPU server inside a private loop: requests and responses never reach the model vendor and are not used for training.
What context window and license does Kimi K2 1T-A32B have?
Kimi K2 1T-A32B holds 32K tokens of context per request and ships under the Modified MIT License license, so you can use it commercially on your own hosting terms.