Serverless GPU inference for open models
GPULoft runs open-weight models on dedicated H100 and H200 capacity behind an OpenAI-compatible API, with bring-your-own-key routing to your own upstream providers and vLLM clusters.
OpenAI-compatible, one line to switch
from openai import OpenAI client = OpenAI(base_url="https://api.gpuloft.com/v1", api_key="YOUR_GPULOFT_KEY")
Bring your own provider keys, route to dedicated vLLM clusters, pay per token.