CobabaAi Features
Unified OpenAI-compatible API
Point any OpenAI SDK or HTTP client to CobabaAi:
text
https://cobabaai.com/v1 (hosted)
https://your-domain.com/v1 (self-hosted)Route to OpenAI, Claude, Gemini, Azure OpenAI, AWS Bedrock, and many other adapters without changing client code.
Channel routing
| Capability | Description |
|---|---|
| Multiple channels | Same model name, different upstream accounts |
| Weight & priority | Control traffic split and preference |
| Auto failover | Switch channel when upstream errors |
| Model mapping | Public model name → upstream model ID |
Billing and quotas
- Per-user balance and usage logs
- Model-level ratios and fixed prices
- Expression-based billing for advanced tiers
- Top-up and redemption flows (when enabled)
Operations
- Full request logging with token counts
- Rate limits (global and per user)
- OAuth sign-in options
- Multi-language admin console
Multi-modal
Depending on enabled channels, CobabaAi can relay chat, images, audio, and video-related APIs supported by upstream providers.
CobabaAi vs generic proxies
CobabaAi is a productized gateway: console, users, tokens, pricing, and logs are built in — not only a thin HTTP proxy.
You still bring your own upstream credentials (on self-hosted) or use CobabaAi's hosted channel configuration.