Privatemode
EU-hosted confidential inference via Cosmian VM
Privatemode delivers privacy-first AI inference from EU data centers using Cosmian VM — a hardened confidential VM combining AMD SEV-SNP and Intel TDX. Accessed via a local proxy binary, the service is designed for GDPR-sensitive workloads where data must remain within a cryptographically isolated enclave throughout processing.
API Features
OpenAI Compatible ✓
Streaming ✓
Function Calling —
Batch Inference —
Vision ✓
Embeddings ✓
Models & Pricing
| Model | Family | Context | Max Out | Input /M | Output /M | Free |
|---|---|---|---|---|---|---|
| GPT-OSS 120B | OpenAI GPT | 128K | — | $5.89 | $5.89 | — |
| Gemma 3 27B | Google Gemma | 128K | — | $5.89 | $5.89 | — |
| Qwen3-Coder 30B-A3B | Qwen | 128K | — | $5.89 | $5.89 | — |
| Qwen3 Embedding 4B | Qwen | 32K | — | $0.15 | $0.00 | — |
| Voxtral Mini 3B | Mistral | 32K | — | €0.004/min audio | — | — |
| Whisper Large v3 | OpenAI GPT | 29K | — | €0.014/min audio | — | — |