Ask anything. Every response comes back with a governance overlay —
claim tiers, source quotes, calibration check, truth budget, and live cost.
Powered by DigitalOcean H100 Serverless Inference.
Mode
ModelPREMIUM
Per-turn — switch anytime. Each governance audit shows which model produced it (expand the audit on any response to see). For a clean head-to-head comparison of the same prompt across models, click 🔄 New conversation between trials so prior context doesn't bias the answer. Pricing for your deployment will mirror these tiers — choose the cost/quality posture that fits your use case.
Add evidence (optional) — required for VERIFIED-tier claims
JSON array of evidence strings. Claims are checked against these spans; without evidence the gate will mark claims UNVERIFIED.
Use sample evidence:
no active thread
Need an idea? Try a sample:
Gateway settings (advanced — leave defaults to use the public demo)