Quickstart

1. Try it with the sandbox key

This shared sandbox key is rate-limited and rotated periodically. Create your own key in the console for production use.

export GPULOFT_API_KEY="gpl_test_pUX4VptdxKJeNWSrvcVieEZlH22dxmbY"

curl https://api.gpuloft.com/v1/chat/completions \
  -H "Authorization: Bearer $GPULOFT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "llama-3.3-70b-instruct", "messages": [{"role": "user", "content": "Hello"}]}'

2. Models

ID	Type	Context
llama-3.3-70b-instruct	chat	131072
qwen2.5-72b-instruct	chat	131072
mistral-small-24b-instruct	chat	32768
deepseek-r1-distill-llama-70b	chat	131072
bge-m3	embedding	8192

3. Bring your own keys (BYOK)

Teams can route traffic through their own upstream provider keys or dedicated vLLM clusters. Manage them in the console or via the API:

curl https://console.gpuloft.com/api/v1/provider-keys \
  -H "Authorization: Bearer $GPULOFT_API_KEY"

4. MCP server

Agents can manage models, usage and provider keys through our MCP server at https://mcp.gpuloft.com/mcp (Streamable HTTP, Bearer auth).

{
  "mcpServers": {
    "gpuloft": {
      "url": "https://mcp.gpuloft.com/mcp",
      "headers": { "Authorization": "Bearer ${GPULOFT_API_KEY}" }
    }
  }
}