Nestor Serve

Deploy a model to Nestor-managed GPU capacity and invoke it through the job API or, for supported language models, the OpenAI-compatible chat API.

Nestor supports both elastic and dedicated operation. Scale-to-zero is a capacity setting, not a different product or API.

Deploy a model

Public Hugging Face text models can be deployed by repository and exact commit. Nestor selects compatible serving software. Customer images and ComfyUI remain available for applications that include code, dependencies, custom nodes or embedded weights.

See Model deployments for deployment and OpenAI-client examples.

Submit a job

export NESTOR_API_KEY=nsk_...
export NESTOR_ENDPOINT=ep_...

curl -sS -X POST \
  "https://api.serve.nestor.software/v1/endpoints/$NESTOR_ENDPOINT/jobs" \
  -H "Authorization: Bearer $NESTOR_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: your-record-id" \
  -d '{"input":{"value":21},"metadata":{"your_id":"12345"}}'

The shape of input is defined by the active runtime. A generic HTTP runtime forwards it as JSON. A ComfyUI runtime expects an API-format workflow.

Poll while integrating:

curl -sS "https://api.serve.nestor.software/v1/jobs/$JOB_ID" \
  -H "Authorization: Bearer $NESTOR_API_KEY"

Use signed webhooks in production. See Integration patterns.