### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### The Feature

## 💡 Proposed Solution

We propose introducing **Vector Multi-Tenancy Configuration** in `config.yaml` and header-based tenant routing inside LiteLLM Proxy.

### 1. Header-Based Tenant Identification Allow LiteLLM Proxy to extract tenant identity from configurable incoming request headers or JWT claims: * `x-tenant-id` / `x-client-id` * Bearer token claims (`tenant_id`)

### 2. Tenant Vector Isolation Modes Support two primary industry-standard multi-tenancy models:

#### Mode A: Index / Collection / Corpus Name Prefixing (Hard Isolation) Automatically partition vector stores and cloud RAG engines by prefixing index, collection, or corpus names with the resolved `tenant_id`: ``` Target Index/Corpus = {tenant_prefix}_{requested_index} Examples: • OpenSearch / PGVector / Qdrant: tenant_52149_clinical_docs_index • Azure AI Search: tenant-52149-clinical-docs-index • Vertex AI RAG Engine: projects/{project}/locations/{region}/ragCorpora/tenant_52149_{corpus_id} ```

#### Mode B: Automatic Metadata Filter Injection (Soft Isolation / Shared Index)…