Model Routing & Providers
Summary
Section titled “Summary”RedCloud is provider-agnostic for AI. You configure one or more providers, and a router directs each request to the right model with failover and a spend budget — so AI features stay available and cost-controlled.
Supported providers
Section titled “Supported providers”| Provider | Notes |
|---|---|
| Anthropic (Claude) | Native support |
| OpenAI | GPT models |
| Google Gemini | Gemini models |
| Google Vertex AI | Gemini via Vertex, Workload Identity supported |
| xAI (Grok) | — |
| Mistral | — |
| DeepSeek | — |
| Ollama | Local models (no data leaves your environment) |
A range of models across these providers is available for selection.
How it works
Section titled “How it works”Routing & failover
Section titled “Routing & failover”The model router directs each request based on configuration, tracks provider health, and fails over to the next provider in the configured order if one is unavailable — so a single provider outage doesn’t break AI features.
Budgets & cost tracking
Section titled “Budgets & cost tracking”A configurable USD budget (overall and per-tenant) bounds AI spend, and cost tracking records usage so you can see and trend AI cost. This keeps AI predictable and prevents runaway expense.
Configuration
Section titled “Configuration”Providers, the default model, fallback order, and budgets are set under Administration → System Settings → AI Configuration and Model Routing. Local models (Ollama) keep inference entirely within your environment when data residency matters.
Notes & limitations
Section titled “Notes & limitations”- AI features may be gated by your license plan.
- For strict data-residency needs, prefer a local provider (Ollama) or Vertex with Workload Identity.