Add support for configurable base_url in autoembedding

[SumJest]

Proposal:

Currently, Manticore Search supports auto embeddings for several providers (OpenAI, Voyage, Jina, Sentence Transformers). For cloud-based providers (like OpenAI, Voyage, Jina), the endpoint is hardcoded to the official API base URL.

Problem: In some environments, users may need to use:

• Custom endpoints (e.g., OpenAI-compatible APIs like Azure OpenAI, local proxy servers, or self-hosted gateways).
• Private/self-hosted inference endpoints that expose OpenAI-compatible or HuggingFace-compatible APIs.

Right now, there is no way to override the default base_url, which makes it impossible to use Manticore’s autoembedding feature with these setups.

Proposal: Add a new optional parameter BASE_URL='<custom_url>' for auto embedding configuration. For example:

CREATE TABLE products_openai (
    title TEXT,
    description TEXT, 
    embedding_vector FLOAT_VECTOR KNN_TYPE='hnsw' HNSW_SIMILARITY='l2'
    MODEL_NAME='openai/text-embedding-ada-002' FROM='title,description' API_KEY='...' BASE_URL='https://custom-endpoint.example.com/v1'
);

Use cases:

• Support for Azure OpenAI Service, which uses endpoints like https://<resource>.openai.azure.com/ instead of https://api.openai.com/.
• Allowing users to connect to self-hosted API gateways or custom providers that expose OpenAI-compatible or HuggingFace-compatible APIs.
• Increased flexibility and easier integration in enterprise environments with strict networking/security requirements.

This change would make auto embedding much more flexible while remaining backward compatible (if BASE_URL is not provided, defaults remain unchanged).

Checklist:

^{To be completed by the assignee. Check off tasks that have been completed or are not applicable.}

• [ ] Implementation completed
• [ ] Tests developed
• [ ] Documentation updated
• [ ] Documentation reviewed
• [x] OpenAPI YAML updated and issue created to rebuild clients

[Sherler]

+1

[anonNo2]

+1