Docusign charges $25/user/month to put a squiggle on a PDF. We charge nothing for specs. Pay only when you run in the cloud — and only for what you use.
Free forever. Run locally with your own API keys.
We run it in the cloud. You own the output.
Deploy on your own infra. We help you get there.
Salesforce charges $300/seat/month. Notion charges $16/seat/month. We charge nothing per seat. Add your whole team. Your whole company. We don't care. Costs scale with usage, not headcount.
| Feature | CLI Free | Managed Runs Starts at $5/run | Enterprise Custom |
|---|---|---|---|
| Running Specs | |||
| Local execution (your machine) | ✓ | ✓ | ✓ |
| Cloud-managed execution | — | ✓ | ✓ |
| Live run logs in browser | — | ✓ | ✓ |
| Bring your own AI API keys | ✓ | Included | ✓ |
| Downloadable output archive | ✓ | ✓ | ✓ |
| Marketplace | |||
| Browse & search all specs | ✓ | ✓ | ✓ |
| Fork & customize specs | ✓ | ✓ | ✓ |
| Publish specs & earn revenue | ✓ | ✓ | ✓ |
| Private spec registry | — | — | ✓ |
| Team & Access | |||
| Number of users | Unlimited | Unlimited | Unlimited |
| SSO / SAML | — | — | ✓ |
| Audit logs | — | — | ✓ |
| Support | |||
| Community Discord | ✓ | ✓ | ✓ |
| Email support | — | ✓ | ✓ |
| Dedicated Slack channel & SLA | — | — | ✓ |
Platform-key managed runs are priced as: max(estimated model cost × 1.4, $5 minimum). If you bring your own API key, we charge a flat $2/run infrastructure fee.
Free open-source CLI, you pay your own model provider directly
We handle cloud execution + model billing
Use your own API key, pay us only infrastructure
We publish conservative, source-backed pricing only. Cost distribution examples will be added once we expose enough aggregated managed-run data.
Because we hate subscriptions too. Monthly fees are how SaaS companies extract value from you whether or not you're using the product. We charge when you run — nothing when you don't. If you run 0 managed runs this month, you pay $0. That's it.
Yes. Full stop. The output of every run belongs to you. MIT-license it, keep it proprietary, sell it — we don't care. We have no rights to anything you build. You own the output forever.
CLI runs execute on your own machine using your own AI API keys. You pay OpenAI/Anthropic directly — we don't touch that cost. Managed runs execute on our infrastructure in a Docker container. We handle the AI billing, you get a single itemised charge from us.
Cost = AI model token usage + compute time (per-second, billed only while your run is active). We add a small platform margin on top. The full cost breakdown — tokens in, tokens out, compute seconds, platform fee — is visible on every run's detail page.
Yes. Every time someone runs your spec as a managed run, you earn a rev-share of the run cost — paid out monthly via Stripe Connect. The better your spec (higher success rate, lower cost), the more runs it gets, the more you earn.
New accounts get one free managed run to try it out. After that, you add a payment method and pay per run. There's no trial timer, no upgrade pressure — just one free run to see what the experience is like.
For managed runs: Claude Sonnet (fast, cheap) and Claude Opus (powerful, more expensive). The spec author recommends a model; you can override it. For CLI runs: whatever model your local setup supports — OpenAI, Anthropic, Ollama, you name it.
Companies that need air-gapped deployments, private spec registries (so your internal specs never hit our public marketplace), custom AI model endpoints (Bedrock, Azure OpenAI, Vertex), and compliance features (SSO, audit logs, SOC 2 artifacts). If you're a regulated industry or just don't want your specs shared publicly, this is for you.
Start with the free open-source CLI. Use managed runs only when you want hosted convenience.