Frequently asked questions
Direct answers for business and technical decision-makers evaluating TechVera.
Is TechVera an AI gateway?
- Cost transparency — per-request EUR cost across providers, attributable per team and application.
- Latency optimisation — per-route, on measured behaviour rather than nominal SKU.
- Governance — scoped API keys, budget enforcement, exportable per-request audit log.
- Sovereignty — EU-operated gateway, EU-contracted entity, per-request jurisdiction policy.
It is not just aggregation; it is the management layer above the providers.
How does TechVera compare to OpenRouter, LiteLLM, Portkey, Helicone or Together AI?
- Gateway operation — TechVera runs the gateway on EU-sovereign cloud (OVHcloud Paris); the US alternatives run on US cloud, so a customer's clear-text prompt traverses a CLOUD-Act / FISA-702 / Patriot-Act exposed surface even when the inference endpoint sits in the EU.
- Contractual jurisdiction — TechVera contracts via an Austrian GmbH under EU law; the US alternatives contract via Delaware Corps, where a DPA cannot cure CLOUD-Act exposure (Schrems II precedent).
- Inference routing — like the others, TechVera routes per request across global, EU-hosted and EU-sovereign providers. This layer alone is replicable; the first two are not.
Beyond the sovereignty structure, TechVera adds cost transparency (per-request EUR across providers), latency optimisation (measured per-route) and governance (scoped keys, audit log, policy enforcement) as first-class management features. The API is OpenAI-compatible, so migration is a base URL change.
Is TechVera GDPR-compliant?
How does TechVera address the EU AI Act and NIS2?
What does "EU sovereign" mean in practice?
Which LLM providers does TechVera support?
- EU Sovereign — Mistral AI (France), Nebius Token Factory (Netherlands HQ; inference in Finland and France), Scaleway (France), IONOS AI Model Hub (Germany), STACKIT (Germany).
- EU Hosted (opt-in, CLOUD-Act labelled) — Azure OpenAI (Sweden Central Data Zone), AWS Bedrock (Stockholm + EU regions).
- Global (opt-in) — Google AI / Gemini (United States).
The full catalogue with every routable model is at /models. The catalogue grows on customer demand — to nominate a provider for the next wave, write to office@techvera.ai.
Is the API OpenAI-compatible?
https://api.techvera.ai/v1. Existing code using the OpenAI SDK migrates by changing the base_url and API key — no other changes required. The model parameter is used to select a routing policy or a specific model.Where is customer data processed?
Who operates TechVera?
The company is 100 % privately owned in Austria — no foreign capital, no parent company outside the EU. The same sovereignty bar we apply to upstream providers applies to TechVera itself.
Contact office@techvera.ai.