diff --git a/pages/docs/configuration/librechat_yaml/ai_endpoints/ai-gateway.mdx b/pages/docs/configuration/librechat_yaml/ai_endpoints/ai-gateway.mdx
new file mode 100644
index 000000000..651dc0ac2
--- /dev/null
+++ b/pages/docs/configuration/librechat_yaml/ai_endpoints/ai-gateway.mdx
@@ -0,0 +1,53 @@
+---
+title: Vercel AI Gateway
+description: Example configuration for Vercel AI Gateway
+---
+
+# [Vercel AI Gateway](https://vercel.com/docs/ai-gateway)
+
+> Vercel AI Gateway: [vercel.com/docs/ai-gateway](https://vercel.com/docs/ai-gateway)
+
+**Notes:**
+
+- **Known:** icon provided, fetching list of models is recommended. Use endpoint name `"Vercel"` in your configuration.
+
+- Vercel AI Gateway provides a unified API for multiple AI providers including OpenAI, Anthropic, Google, and others through a single endpoint.
+
+- `stop` is recommended to be dropped as different underlying models use different stop tokens.
+
+- **Reasoning models:** AI Gateway supports extended thinking via the `reasoning` parameter with `enabled`, `effort` (none/minimal/low/medium/high/xhigh), and `max_tokens` options. Response includes `message.reasoning` with the thinking content.
+
+```yaml
+    - name: "Vercel"
+      # For `apiKey` and `baseURL`, you can use environment variables that you define.
+      # recommended environment variables:
+      apiKey: "${AI_GATEWAY_API_KEY}"
+      baseURL: "https://ai-gateway.vercel.sh/v1"
+      titleConvo: true
+      models:
+        default:
+          - "openai/gpt-5-mini"
+          - "google/gemini-2.5-flash"
+          - "deepseek/deepseek-v3.1"
+          - "anthropic/claude-3-haiku"
+        fetch: true
+      titleModel: "openai/gpt-5-mini"
+```
+
+## Environment Variables
+
+Add the following to your `.env` file:
+
+```bash
+AI_GATEWAY_API_KEY=your_api_key_here
+```
+
+## Supported Features
+
+| Feature | Notes |
+|---------|-------|
+| Model fetching | Via `/v1/models` endpoint |
+| Streaming | Full SSE support |
+| Vision/Images | Depends on underlying model |
+| Reasoning | Via `reasoning` parameter with `enabled` and `effort` |
+| Tool calling | OpenAI-compatible function calling format |