[Feature]: DeepInfra provider for Qwen3-reranker

[qdrddr]

Use Case

I'd like to use the DeepInfra provider with the SOTA Qwen3-reranker model

Problem Statement

It's cheaper than Cohere and has great quality

How This Feature Would Help

Reduce costs

Proposed Solution

Implement LiteLLM Client SDK so I could easily switch between providers and models

Alternatives Considered

No response

Priority

Important - affects my workflow

Additional Context

No response

Checklist

• I would be willing to contribute this feature

[nicoloboschi]

@qdrddr we already support litellm rerankers, can you try it out ?
https://hindsight.vectorize.io/developer/models#litellm-supported-providers

[qdrddr]

@qdrddr we already support litellm rerankers, can you try it out? https://hindsight.vectorize.io/developer/models#litellm-supported-providers

Currently, the way it is implemented requires running LiteLLM Proxy.

While the LiteLLM Client SDK could handle other models & providers, including LLM, Embeddings & Reranker models, out of the box without the need for the proxy.

My request was to implement and use the LiteLLM Client SDK (not to be confused with the LiteLLM Proxy). @nicoloboschi