Custom Provider Base URLs Tracking Issue

This issue tracks requests for custom API endpoint support across different model providers.

Related Issues

By Provider

• [Feature Request] Support for Custom Anthropic-Compatible API Endpoints #3162 - Custom Anthropic-compatible API endpoints (crashes with mapOptionsForApi error)
• Feature Request: Custom baseUrl for model providers (OpenAI-compatible proxies) #2305 - Custom baseUrl for OpenAI provider (LiteLLM proxy support)
• [Feature]: Need an easy way to change the llm endpoint. #2610 - General request for easier LLM endpoint configuration in UI

Related TTS Issue

• Custom TTS endpoint configuration is difficult - silent validation failures and undocumented workaround #2619 - Custom TTS endpoint configuration (ElevenLabs baseUrl)

Use Cases

1. Internal LLM gateways — Cost tracking, security, compliance, audit logging
2. Local proxies — LiteLLM for smart routing (Ollama → Gemini → Claude fallback)
3. Self-hosted models — Route to local inference servers
4. Rate limit management — Distribute load across multiple endpoints

Current State

Proposed Solution

1. All providers should support baseUrl in config
2. API type detection should not assume official endpoints
3. Validation should be relaxed when custom baseUrl is configured
4. Document which features work with custom endpoints

Acceptance Criteria

• models.providers.<provider>.baseUrl works for all built-in providers
• No crashes when using custom endpoints
• Clear documentation for proxy/gateway setups