Guarantee schema‑true outputs across models to cut parsing errors

We introduced a strict_output setting that enforces exact adherence to your output_schema by default across supported models. This delivers consistent, machine-parseable responses without extra validation layers, reducing integration effort and downstream failures. You can relax enforcement when needed by switching to guided mode.

Details

Strict mode is on by default to return outputs that match your schema exactly
Guided mode offers more flexibility when strictness isn’t required
Fewer retries, simpler validators, and more reliable pipelines

Who this is for: Teams building structured workflows, data pipelines, or API integrations that require deterministic, schema-compliant outputs.