Self-serve
$0.20per agent-hour on a Small machine
Each agent is billed separately, based on actual agent hours used, at the rate of the machine size you pick. Bring your own keys for 300+ model providers, or use the optimised models on the ForgeCode gateway at API pricing.
| Machine | Per agent-hour |
|---|---|
| Small 2 vCPU / 8 GB RAM / 128 GB disk | $0.20 |
| Medium 4 vCPU / 16 GB RAM / 256 GB disk | $0.40 |
| Large 8 vCPU / 32 GB RAM / 512 GB disk | $0.80 |
| Ultra 16 vCPU / 64 GB RAM / 1024 GB disk | $1.60 |
For your organisation
Custom
Priced case by case. Everything in pay as you go, plus your own cloud, organisations and teams, SAML SSO, custom usage limits and a forward-deployed engineer.
Bring your own cloud
AWS, Google Cloud or AzureCompare plans
| Pay as you go | Enterprise | |
|---|---|---|
| Models | ||
| 300+ model providers Bring your own keys, out of the box | Both plans | |
| ForgeCode gateway A few hyper-optimised models, no keys needed | Both plans: pay as you go at API pricing from prepaid credit | |
| Infrastructure | ||
| Cloud factories On Forge-managed infrastructure | Included | Included |
| Bring your own cloud | —Not included | Included: AWS, Google Cloud or Azure |
| Organisations and teams Multiple people work together on shared factories and mission boards | —Not included | Included |
| SAML SSO Sign in through your identity provider | —Not included | Included |
| Usage | ||
| Automations Triggers and watchers that start, nudge, re-run and escalate work | Limited | Custom |
| Artifacts Files agents produce, such as video proof and test reports | Limited | Custom |
| Knowledge What the factory remembers about your codebase and incidents | Limited | Custom |
| Reports Live dashboards and executive summaries of where work stands | Limited | Custom |
| Crons Work that runs on a schedule | Limited | Custom |
| Cloud machines The machines your factories run on | Limited | Custom |
| Support | ||
| Community support | Included | Included |
| Forward-deployed engineer Works with you on your business domain, understands your requirements, sets up the factory for your use case and keeps optimising it to deliver your goals | —Not included | Included |
Frequently asked questions
What is an agent-hour?
One hour of one agent running in a factory. Each agent is billed separately, based on actual agent hours used, at the hourly rate of the factory’s machine size: $0.20 on Small, $0.40 on Medium, $0.80 on Large and $1.60 on Ultra. For example, three agents running for an hour on a Small machine cost $0.60.
Which machine size do I need?
Start with Small. Move a factory to Medium, Large or Ultra when its builds, test suites or services need more CPU, memory or disk. Each size up doubles the vCPUs, memory, disk and hourly rate.
How is model usage billed?
Two ways. Bring your own keys for any of 300+ supported providers, and your provider bills you directly. Or use the few optimised models on the ForgeCode gateway, pay as you go at API pricing from prepaid credit.
Which models can I bring?
ForgeCode supports 300+ model providers out of the box. Bring your own keys for the provider you want to use.
What is the ForgeCode gateway?
The gateway gives you a few hyper-optimised models that work really well with ForgeCode, without bringing your own keys. Usage is pay as you go at API pricing from prepaid credit.
Can I run factories in my own cloud?
Yes, on Enterprise. Bring Your Own Cloud runs factories in your own AWS, Google Cloud or Azure account. Book a demo to set it up.
What is a forward-deployed engineer?
Part of Enterprise support: an engineer from the ForgeCode team who works with you on your business domain. They understand your requirements, set up the factory for your use case, and keep working with you to optimise it so it delivers your goals.
Do you train AI models on my data?
No. We do not use your content, including your code, files, prompts, conversations and outputs, to train AI models. Prompts and context sent to the model providers you choose are handled under those providers’ own terms, so check their data and training policies.