# Does a custom classifier cost more to run than a built-in one?

No. A custom classifier is built to run on the same terms as the built-ins: cheap enough to score every trace instead of a sample, so the per-event cost stays negligible whether the classifier is generic or fine-tuned to your agent. Training it costs more upfront, since it needs labeled examples from your own traffic before it ships, but that's a one-time cost, not a per-trace one.

The reason a custom classifier can afford to run on everything is the same reason a built-in can: it scores one narrow property at low cost. That's what makes checking every trace affordable instead of a tradeoff against coverage, whether the classifier is generic or fine-tuned to your agent.

---

Source: https://tessary.ai/answers/tessary-custom-classifiers/does-a-custom-classifier-cost-more-than-a-built-in-one
More on Tessary custom classifiers: https://tessary.ai/answers/tessary-custom-classifiers
From Tessary, agent reliability for AI agents in production: https://tessary.ai