feat: add Ollama rerank model support

[timon0305]

Summary

Adds rerank model support for the Ollama provider, enabling users to use Ollama-hosted reranking models (e.g., via llama.cpp's /v1/rerank endpoint or compatible servers).

Changes

• rag/llm/rerank_model.py — Added OllamaRerank class implementing the Base rerank interface using Ollama's /v1/rerank HTTP endpoint. Supports optional API key authentication and follows the same pattern as XInferenceRerank and LocalAIRerank.
• conf/llm_factories.json — Added TEXT RE-RANK to the Ollama factory's tags so it appears as a rerank-capable provider in the UI.
• web/src/pages/user-setting/setting-model/modal/ollama-modal/index.tsx — Added explicit LLMFactory.Ollama entry to the optionsMap with rerank as a supported model type, rather than relying on the Default fallback.
• docs/references/supported_models.mdx — Added rerank checkmark for Ollama in the supported models table.

How it works

The OllamaRerank class uses the /v1/rerank API endpoint (compatible with llama.cpp server, Ollama community forks, and other OpenAI-compatible rerank servers). It sends a POST request with the query and documents, and parses the results array containing index and relevance_score fields.

Closes #4406
Closes #3680

[dosubot]

Related Documentation

Checked 0 published document(s) in 1 knowledge base(s). No updates required.

^{How did I do? Any feedback?}