Planned pricing

One end-to-end run: prompt → manifest → code generation → build → verify → artifact. If verification fails and the system auto-retries internally, that still counts as one job. Cancelled jobs aren't billed.

One round-trip to a model provider through our gateway — including planner calls, codegen calls, repair calls, and interview chat turns. We expose per-call cost in the Trace tab so you can see exactly where your quota went.

No. The managed multi-model gateway is shared across tiers. BYO keys are supported today via /admin/factory/settings/api-keys — provider bills then flow directly to your Anthropic / OpenAI / etc. account. Pro / Team / Enterprise tier-specific BYO gating is planned.

Soft cap at 80% (email warning) and hard cap at 100% (jobs queue rather than overage-billing) are the planned Pro+ model — no quota system is wired today because Pro / Team / Enterprise haven't shipped. Auto-upgrade is on the roadmap for the Settings → Billing surface (which itself is planned).

Bridge minutes are wall-clock time spent on the Linux build node during code generation, compile-fix, and launch verification. A typical web build takes 4-12 bridge minutes including npm install, next build, server launch, and the POST + GET smoke. The desktop / mobile bridges are roadmap (per ADR 0008) — no bridge-minute cost is quoted there until they ship verified runs.

Hobby tier is free with no commitment. Pro / Team / Enterprise tiers are planned — no billing system is wired yet, so there is nothing to cancel from. When billing ships, the intended model is month-to-month with no early-termination fee.

There is no paid trial because the paid tiers don't ship yet — Pro / Team / Enterprise are planned. Hobby tier is free and usable today with no credit card. When paid tiers ship the intended model is a 14-day full-access trial; until then, the public demo at /factory#demo and the Hobby tier are the entry points.

Not yet a formal program. Hobby tier is free today. When paid tiers ship, a 50%-off path for verified education / nonprofits / OSS is on the roadmap — write to [email protected] if you fit one of these segments today and we will track you as a launch customer.

No. Our DPA at /factory/dpa governs data use; OpenAI and Anthropic Zero-Data-Retention enrollment is in progress on our gateway path. Your prompts and generated code are never used for training. Prompts and outputs are retained only for the artifact retention window — see /factory/privacy.

Not yet. A first-party docker-compose / Terraform self-host story is on the roadmap (ADR 0008). The generated workspaces ARE plain Next.js + libSQL — you can host those anywhere Next.js runs (Vercel, Cloudflare Pages, Docker, your own server) — but the factory itself does not yet ship a self-host bundle.

Response-time commitments (Pro within 4 business hours, Team within 1 business hour, Enterprise 30 minutes 24/7) are targets on planned tiers — no contract is signable today because billing is not wired. On Hobby (the only live tier), support is best-effort via email and GitHub Discussions.

99.9% uptime SLA on Team and Enterprise tiers (service credits for breaches) — these tiers are planned, not signable today. Hobby and Pro are best-effort. /factory/status shows live component status (bridges, API, proof page). Automated 90-day uptime tracking is on the roadmap; we do not yet publish a measured-uptime number.