SwarmDo GPU — One API key for GPUs, LLMs and video generation

First call in five minutes

Create an account, add credit, copy a key from the dashboard. Then any of these works.

Hosted model · curl

Generate a 5-second video

One request, one clip. The charge for the call comes back in the x-swarmdo-charged-cents header.

curl -X POST https://gpu.swarmdo.com/api/managed/kling-2-1-pro/runsync \
  -H "Authorization: Bearer $SWARMDO_API_KEY" \
  -H "content-type: application/json" \
  -d '{"input":{"image":"https://…/frame.png","prompt":"slow dolly in","duration":5}}'

Serverless · OpenAI-compatible

Point your existing client at us

Deploy a model to an endpoint from the dashboard, then swap the base URL. Cursor, Continue, Open WebUI and the OpenAI SDKs work unchanged.

# python
client = OpenAI(
  base_url="https://gpu.swarmdo.com/run/<endpoint-id>/openai/v1",
  api_key=os.environ["SWARMDO_API_KEY"])

Agents · MCP & SDK

Let an agent spend the credit

44 tools: rent GPUs, deploy endpoints, generate video, check the bill.

# hosted — nothing to install
claude mcp add --transport http swarmdo-gpu https://gpu.swarmdo.com/mcp \
  --header "Authorization: Bearer swdo_…"

# or local
claude mcp add swarmdo-gpu --env SWARMDO_API_KEY=swdo_… -- npx -y @swarmdo/gpu-mcp
npm i @swarmdo/gpu-sdk   # in code

Built for teams that resell AI work

Agencies and studios get one consolidated bill and the controls to hand it to a client.

SquadsPool credit, allocate it per project, see who spent what.

Budgets & forecastsMonthly budget with a burn-rate forecast and alerts before you hit it.

Automatic top-upA saved card refills the balance so running work is never stopped for credit.

Audit logEvery provision, call and charge, exportable.

SLA tiersUptime targets with automatic credits when they are missed.

Custom domainsServe an endpoint from your own hostname.

Pricing

Per-second billing on prepaid credit. Minimum top-up $5, no subscription, no idle charge on scale-to-zero endpoints.

Serverless GPU endpoints

Loading…

Hosted models

Loading…

GPU rentals are priced per offer in the catalog below. Serverless is billed per GPU-second while a request is running; hosted models are billed per clip, per request or per million tokens, and only when the call succeeds.

One API key for GPUs, LLMs and video generation.

Three ways to buy compute, one balance