Full reference

GPU offers and instances

Find available capacity, launch an instance with explicit placement and access settings, then inspect and terminate it when the work is complete.

Updated

Reviewed against Console v0.2.2. Available actions depend on project access and runtime configuration; check your installed CLI version for command differences.

Inspect compute before provisioning

Open Settings → Organisation → Compute to inspect offers, jobs, and instances. Models shows per-model assignment/readiness. A listed GPU offer is not a reservation or proof the account is admitted for provisioning. If you already own suitable hardware, use Pair local compute instead of renting a new instance.

Inspect current offers

bash
stimulir compute offers --json

Choose an offer and region from the current response. Provider credentials, available capacity and the configured runtime determine which offers can be returned. An empty or unavailable provider result does not establish that other providers have no capacity.

Launch with the required inputs

bash
stimulir compute up <offer-id> \
  --region <offered-region> \
  --ssh-key <registered-key-name> \
  --count 1

The region and provider-registered SSH key are required. Use --provider when the offer is from a provider other than the command’s default. Provisioning can incur charges; a returned instance ID may still be pending.

Inspect and terminate

bash
stimulir compute list
stimulir compute status <instance-id>

Check readiness before sending a workload. When finished, run stimulir compute down <instance-id> and verify termination. Stopping a training job and terminating an instance are different operations.

Choose the endpoint deliberately

The default command path goes through the configured Console service to its runtime. --endpoint targets a runtime directly. Use Workers when you need a bootstrapped training worker rather than a raw instance.

Command catalogue

Expand a command for its arguments. These commands follow the released CLI source; check stimulir --version and the command’s --help for your installation. Required flags are marked; some operations also validate combinations of inputs.

stimulir compute down

Terminate an instance.

instance_id required
Positional argument. Instance id.
--yes, -y
Skip the confirmation prompt.
--endpoint
HybrIE runtime URL (bypasses the console proxy).
--json
Emit raw JSON output.
stimulir compute list

List compute instances.

--endpoint
HybrIE runtime URL (bypasses the console proxy).
--json
Emit raw JSON output.
stimulir compute offers

List available GPU offers.

--provider
Compute provider. Omit to list every reachable provider.
--region
Only offers available in this region.
--endpoint
HybrIE runtime URL (bypasses the console proxy).
--json
Emit raw JSON output.
stimulir compute status

Show a single instance.

instance_id required
Positional argument. Instance id.
--endpoint
HybrIE runtime URL (bypasses the console proxy).
--json
Emit raw JSON output.
stimulir compute up

Launch instance(s) from an offer.

offer_id required
Positional argument. Offer id (from stimulir compute offers).
--region required
Region to launch in.
--ssh-key required
SSH key name registered with the provider (repeatable).
--count
Number of instances.
--provider
Compute provider.
--spot
Request spot capacity.
--name
Instance name/tag.
--endpoint
HybrIE runtime URL (bypasses the console proxy).
--json
Emit raw JSON output.

SDK method catalogue

Expand a client to inspect public methods and parameter names. Read each operation’s task guide before supplying a payload or starting work. The SDK returns platform responses; handle errors and inspect completion state before using a result.

client.computeComputeClient
offers(provider, region)
Purchasable GPU configurations.
offer_sources()
Which providers answered the aggregate listing, and why any didn't.
launch(offer_id, region, ssh_key_names, provider, count, use_spot, image, cloud_init, name, extra)
Launch instance(s) from an offer.
launch_worker(offer_id, region, ssh_key_names, model_repo, model_slug, hybrie_ref, provider, count, use_spot, env)
Provision, bootstrap and register a worker from one exact offer.
instances(provider)
Every instance the provider reports, as {"id", "status"} rows.
status(instance_id, provider)
Returns the platform response; inspect its status and error detail.
terminate(instance_id, provider)
Tear an instance down. Idempotent — an already-gone id is success.
peers()
Remote runtime peers currently registered with the control node.
ensure_warm()
Ask the autoscaler to bring the mesh up to its warm floor.