Use case
How to test a cheaper model without rewriting your application
A cheaper model is only a saving if it still does the job. Reroute rules let you test that without changing application code.
A reroute rule you configure in Spendline can send chosen calls to a different model, so your application keeps calling the same endpoint. Validate task quality on the new model before relying on it; a lower price does not prove equal quality, and no saving is guaranteed.
How it works
Reroute rules are resolved in the request path before the budget check, so the budget prices the model that will actually serve the call. The optimizer view lists lower-priced models alongside your recorded usage; its quality scores are catalogue estimates, not benchmarks run on your tasks.
{
"from_model": "gpt-5.2",
"to_model": "gpt-5.2-mini",
"scope": { "agent_id": "ticket-triage" }
}What to know before relying on it
Spendline does not measure answer quality. A model change can lower cost per token and still raise total cost if the new model needs more tokens or more retries, so compare quality and total cost on your own traffic before keeping a rule.
Why it matters
Changing a model in code means finding every call site and shipping a deploy. A rule in the request path can be tried on one workflow and switched off again without touching application code.
Frequently asked questions
Does rerouting risk degrading quality?
Yes, if you do not test it. A reroute is a rule you configure deliberately for a specific scope; check the outputs on real tasks before widening it.
Can a reroute move traffic to another provider?
Reroute rules work across the providers you route through Spendline, within the supported request shapes.
See what this would look like in your product.
Start with a guided 30-day free pilot on live production traffic. The 30 days start when your traffic flows through Spendline, not at sign-up. We agree the scope, success criteria and paid terms before you start. There is no Spendline subscription fee during the pilot; your model-provider charges still apply.
Prefer email? fida@spendline.ai