Free
For trying a policy on real traffic.
$0/mo
$0.00 / run
10K runs / month
- 1 policy, BYO keys
- Full replayable traces
- 7-day trace history
- Community support
BYO keys · per-run metering
Priced per run, not per token.
unhardcoded does not resell tokens. Your model usage stays with your provider accounts. We charge for the routing layer — priced per run, not per token.
unhardcoded · per-run meter
req-204815
Baseline
your current model, same floor
$0.063/run
On unhardcoded
cheapest model that passes
$0.018/run
savings measured per run · cheapest that meets the rules
↓71%
Illustrative. Real per-run figures appear in your dashboard from day one, against your own baseline.
$$Plans
One line item: runs.
Start free, keep your provider keys, and pay per run.
Monthly
Annual
Save 2 months
WHAT COUNTS AS A RUN?
One chat-completion request is one billable run. A workflow is a multi-step policy chain (each step is a node) — one workflow is one billable run up to 5 nodes. Larger workflows are available on Scale and Enterprise.
Recommended
Pro
For apps routing production AI calls.
$49/mobilled annually
then $0.20 / 1K runs
250K runs / month included
- Unlimited policies & workflows
- Replayable, auditable traces
- Quality floors & price ceilings
- 90-day history · email support
Scale
For teams with high-volume AI traffic.
$199/mobilled annually
then $0.12 / 1K runs
1.5M runs / month included
- Everything in Pro
- Roles, SSO & audit log
- Shared policy library
- 1-year trace history
- Priority support
Enterprise
For compliance, scale, and deployment control.
Custom
volume run rate
Unlimited runs
- Audit-ready trace export
- Custom trace retention
- Self-hosted open-source layer
- Dedicated routing & support
- Named solutions engineer
All figures illustrative · priced per run, not per token · a workflow = one billable run up to 5 nodes · keep your provider accounts
÷Two bills
Two bills, cleanly separated.
Your model usage stays with your provider accounts. We charge for the routing layer. Nothing is bundled, marked up, or resold.
paid to your providers
Your provider bill
You keep your OpenAI, Anthropic, Google, or Mistral accounts and your own keys. They bill you directly for inference. unhardcoded never resells those tokens.
paid to unhardcoded
Your unhardcoded bill
One line item, priced per run: policy execution, model catalogs, traces, replay, health checks, and managed infrastructure — failover, history, team controls, and uptime.
FAQbefore you wire it in
Pricing, answered plainly.
Are you reselling tokens?
No. You bring your provider accounts and pay providers directly for inference. unhardcoded charges only for the managed host.
Does a workflow count as multiple runs?
A workflow counts as one billable run up to 5 nodes. Larger workflow limits are available on Scale and Enterprise.
What if I exceed my plan?
You move to the overage rate for your plan. Pro overage is $0.20 per 1K runs. Scale overage is $0.12 per 1K runs.
Can I self-host instead?
Yes. Self-hosting the open source layer has no per-run fee, but you operate everything yourself: provider modules and their API changes, live catalog and pricing upkeep, key management, failover, uptime, and trace storage. The managed host runs all of that for you for the per-run price.
Why pay if the core is open source?
The open core gives you the policy language and interpreter. The managed host gives you: provider modules, key management, OAuth, live catalog data, traces, history, failover, uptime, and support.
One host. One line item.
Start free, keep your provider keys, and pay per run for the managed host you would otherwise build yourself. Every routing decision is traced and replayable.
No credit cardKeep your provider accountsPriced per run, not per token