Pick the plan by the controls you need and cap usage per person before seats multiply. Route routine work to cheaper models, and decide on expansion from merged work and reviewer time per unit of spend.
When you need this
- A pilot group wants Factory for their teams, and finance wants a monthly figure.
- Engineers keep hitting rate limits, or Extra Usage spend is climbing.
- You need evidence beyond seat counts before you expand.
Evaluation criteria
Plan
Which plan carries the controls the rollout needs?
Look forBudget US$60 per month plus US$40 per seat for Teams, with up to 10 seats billed month to month at Pro rate limits. Business adds SSO, SAML and SCIM provisioning, zero data retention, audit logging, and admin controls for models, autonomy, deny lists, and network policy. Enterprise adds unlimited seats, dedicated compute, on-premise options, sub-organizations, and data residency.
Usage model
How does a seat run out, and what happens then?
Look forTrack all three rolling limits, 5-hour, 7-day, and 30-day. Each request needs room under all of them at the selected model's rate. At a limit, use Droid Core's open-weight models on a separate limit or prepaid Extra Usage. Missions pause at a limit and need Extra Usage enabled. Business and Enterprise replace these limits with contract limits.
Model routing
Which models should routine work use?
Look forRestrict the model catalog and make cheaper models the default. Reserve frontier models for design and complex refactors, as Factory advises. Factory Router selects a model per task. Factory reports a 63% aggregate saving against frontier-model rates, so check that figure against your own Analytics data.
Per-person caps
What stops one engineer or automation from using the whole budget?
Look forOn Enterprise, set monthly per-member limits through User Groups tied to identity provider groups. Add individual overrides where needed. Review consumption by model and user in Factory Analytics. The Cost Concentration chart shows how much of that spend comes from a few power users and automations.
Measurement
How will you know the pilot paid off?
Look forTrack merged pull requests per team and repository through OpenTelemetry exports or the Analytics API. Record reviewer time for each Droid change and the share merged without rework. Use Agent Effectiveness to compare spend with cycle-time changes on issues and pull requests Droid touched.
Pilot scope
Which repositories and tasks does the pilot cover?
Look forPick two or three repositories at Agent Readiness Level 3 or above, with tests that verify changes. Include bug fixes and refactors plus one bounded migration. Reserve Missions for the migration. Enable Extra Usage only for its operators.
Questions to ask vendors
- Which models did each team use, and what share of consumption did each take?
- How often did engineers hit a rate limit, and did they fall back to Droid Core or buy Extra Usage?
- How long did reviewers spend on Droid changes compared with their teammates'?
- Which teams would lose their workflow if we moved from Teams to Business limits?
Red flags
- Expansion decided on seat activity alone.
- Extra Usage enabled for everyone with no owner watching spend.
- Frontier models as the default for every task.
- Missions run on Pro seats without a budget for the Extra Usage they need.
Products in this piece
Sources
- Official documentation
- 11
- Factory: Pricing
- Factory docs: Individual Plans
- Factory docs: Organization Plans
- Factory docs: Available Models
- Factory docs: Factory Router
- Factory docs: Cost & Productivity
- Factory docs: User Groups
- Factory docs: Telemetry & Analytics
- Factory news: Factory Analytics brings transparency to every session (Oct 2, 2026)
- Factory docs: Agent Readiness
- Factory docs: Factory Missions