Getting Started

Quickstart

Go from zero to your first inference call and first training job in about five minutes.

1. Install the CLI

The stimulir CLI is distributed from the Stimulir Console platform and installed with uv:

bash
uv tool install stimulir

2. Log in

Log in with your Stimulir account email and password. The CLI stores your session token in ~/.stimulir/:

bash
stimulir login

3. Select a workspace

Every resource — keys, usage, training jobs — is scoped to a workspace (business profile). List your workspaces and select one:

bash
stimulir workspace list
stimulir workspace use <workspace-id>

4. Create an API key

Inference requests authenticate with hyb_* API keys. Create one for your environment:

bash
stimulir keys create --name "my-first-key" --env dev

The plaintext hyb_* key is shown once at creation. Store it securely — it cannot be retrieved again. See API Keys.

5. Make your first inference call

From the CLI:

bash
stimulir infer chat "Summarise our Q3 churn drivers" --model qwen3-4b --stream

Or hit the OpenAI-compatible endpoint directly with your key:

curl
curl https://api.stimulir.com/api/v1/inference/chat/completions \
  -H "Authorization: Bearer hyb_..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3-4b",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Requests are routed by model prefix to your BYOK credentials or to Managed Inference, and usage is metered per request. See Inference API.

6. Run your first training job

Train a classical PEFT LoRA adapter with supervised fine-tuning on the Lab workspace, then follow its progress:

bash
stimulir lab train sft --family qwen3-4b --lora-rank 16 --epochs 3

stimulir lab jobs list
stimulir lab jobs get <job-id>

Training runs asynchronously with checkpoints; job status includes step and loss progress points. See PEFT Tuning (LoRA) and Training Jobs.

What's next