Model-agnostic · Harness-agnostic
Fine-grained control over your AI spend — in every harness.
Caliper shows you exactly where every dollar goes, then routes the work to cheaper American-hosted open models — right inside Claude Code. Next: custom models fine-tuned on your own usage.
Early access · no spam · ships as a skill you already know how to install.
See how it works- feature-build$128
- app-deploy$94
- pr-review$71
- bug-fix$52
- test-suite$33
Swap to open models → save ~$290 (4.7× cheaper)
The problem
You have no control over what your AI costs.
Powerful new features in Claude Code and harness-level workflows drive far more productivity — but they also explode costs in a way you can't see or steer. You're at the mercy of external pricing on models you don't control. Caliper lets you keep the productivity and take back control of the bill, so you give up nothing.
How it works
See it. Route it. Own it.
Control your spend.
Every token, broken down by workflow and model — cache-aware, not a guess. Know exactly where the money goes and set the limits.
Route your tokens.
Harness-agnostic smart routing sends each task to cheaper American-hosted open models — safely, so you can evaluate before you commit. Your agents become cost-aware and start optimizing their own spend.
Own your model.
We handle the entire fine-tuning process for you — no work on your end. You get a custom model trained on your workload, and once it beats the alternative, your agent automatically starts using it.
Why Caliper
Control your cost.
Surface spend and cut it. See the savings before you switch a single model.
Harness- & model-agnostic.
Works in Claude Code, Codex, Cursor — routes across any model. No new IDE, no lock-in.
Open models, trained on your usage.
The end state: a model that's yours. Personal, or trained on your org's entire body of work.
The difference
Neutral by design.
Your model provider will never route you to a cheaper competitor. A single-harness tool only helps if you live in their harness. Caliper is neither — it's a tool your agent uses to manage its own spend wisely. We make agents cost- and efficiency-aware, so they're always seeking the best performance per dollar, and the leverage stays yours.
For teams
Bring it to your org.
We'll email you when your seat is ready. Refer a teammate and you both get a free month of Pro at launch.
Centralized cost visibility
See spend across every developer, workflow, and model in one place.
A model trained on your team
Your proprietary way of building — not a generic base model everyone else uses.
Your traces stay yours
Privacy and control by default. Training on your usage is opt-in.
Stop overpaying for tokens.
Get early access to Caliper.
No spam, ever.
FAQ