Legal
Last updated: 19 March 2026
Cailos routes your API requests to third-party LLM providers ("subprocessors") based on your routing configuration. The content of your requests is processed by these providers according to their own privacy policies and terms of service.
Before any request reaches a subprocessor, Cailos applies PII shielding (via LLMShield) to detect and cloak personally identifiable information in transit. However, the semantic content of your prompts is visible to the routed provider.
Your requests may be routed to any of the following providers depending on your model selection, optimisation strategy, and endpoint availability:
| Provider | Purpose | Location | Privacy Policy |
|---|---|---|---|
| OpenAI | LLM inference | US | Link |
| Anthropic | LLM inference | US | Link |
| Google (Gemini) | LLM inference | US | Link |
| Cohere | LLM inference | CA | Link |
| Groq | LLM inference | US | Link |
| Together AI | LLM inference | US | Link |
| DeepInfra | LLM inference | US | Link |
| Scaleway | LLM inference | FR (EU) | Link |
| Cerebras | LLM inference | US | Link |
| NCompass | LLM inference | US | Link |
| NScale | LLM inference | EU | Link |
| Inception | LLM inference | AE | Link |
| OpenRouter | LLM inference routing | US | Link |
| Novita AI | LLM inference | US | Link |
| SambaNova | LLM inference | US | Link |
| Tinfoil | LLM inference (secure enclave) | US | Link |
| xAI | LLM inference | US | Link |
| BytePlus | LLM inference | SG (AP-Southeast) | Link |
| Provider | Purpose | Location |
|---|---|---|
| Railway | Application hosting (web server, Celery workers) | US |
| Neon / PostgreSQL | Database (account data, request logs, endpoint config) | US |
| Redis (Upstash) | Caching (rate limits, circuit breakers, live TPS/latency) | US |
| Cloudflare | DNS, CDN, DDoS protection | Global |
| Axiom | Structured log ingestion and observability | EU |
Cailos uses LLMShield to detect and cloak PII before requests reach any LLM inference provider. This means personally identifiable information (names, emails, phone numbers, addresses, etc.) is replaced with tokens before transit and restored in the response. The upstream provider never sees the raw PII.
Each endpoint in Cailos is assigned a trust level (0-3) based on the provider's data handling practices:
You can control which trust levels your requests are routed to via the trust_level routing hint.
We will update this page when subprocessors are added or removed. Material changes will be communicated to account holders via email.