Docs

Launch your first GPU instance

Three steps to your first request. Full API reference and guides are on the way - this page covers the essentials.

Install the CLI

$ curl -fsSL https://get.meridian.finimble.com | sh

Set your API key

Grab a key from your account and export it:
```
$ export MERIDIAN_API_KEY="mk_live_..."
```

Launch an instance or call the API

Start a GPU instance, or hit the inference endpoint directly:

$ meridian instances create --gpu hgx-h100 --image cuda-12.4

$ curl https://api.meridian.finimble.com/v1/chat/completions \
  -H "Authorization: Bearer $MERIDIAN_API_KEY" \
  -d '{"model": "llama-3.1-70b", "messages": [{"role": "user", "content": "Hello"}]}'

Guides

Go deeper

API reference

Endpoints, authentication, rate limits, and error codes for the inference and instance APIs.

Managed Kubernetes

Create GPU-aware clusters, schedule multi-node training jobs, and manage node pools.

Networking

InfiniBand topology, private networking, and firewall rules for your instances.

Storage

Local NVMe, persistent volumes, and moving checkpoints and datasets efficiently.

Images and environments

Prebuilt CUDA images, custom containers, and snapshotting configured environments.

Service status

Live availability and incident history for the Meridian platform.

Questions the docs don't answer yet?

Our full documentation is in progress. In the meantime, the team is happy to help directly.