Frequently asked questions

No. Nexora is built for application engineers, not ML researchers. You compose agents with a typed SDK in the language you already use, and the platform handles model routing, retries, and scaling. Teams with ML expertise can still drop down to fine-tuning and custom evals when they need to.

The SDK ships first-class libraries for TypeScript and Python, with a stable REST and streaming API for everything else. Community clients cover Go, Rust, and Java, and every endpoint is documented with copy-paste examples for each.

Yes. The free tier includes enough monthly requests to build and ship a first agent — no credit card required. You keep full access to tracing and evaluations so you can validate quality before you scale.

You pay for what you use, metered per request and per token across inference, guardrails, and evaluations. Usage is aggregated on a live dashboard, and you can set hard spend limits and alerts per project so there are no surprises at the end of the month.

Absolutely. Upgrade, downgrade, or cancel at any time from the billing settings — changes take effect immediately and we prorate the difference. There are no lock-in contracts on our standard plans.

Yes. Nexora is SOC 2 Type II certified, and our latest report is available under NDA on request. We also support GDPR and offer HIPAA-ready configurations for teams handling regulated workloads.

Data is stored in the region you choose at project creation, across US, EU, and APAC. Data stays resident in that region, is encrypted in transit and at rest, and never leaves it without your explicit configuration.

No. We never train models on your prompts, completions, or traces. Your data is used only to serve your requests, and you control retention windows — set them to zero for ephemeral processing if you prefer.