Update provider examples with zhipu and gemini variants

Author: Makiras

doc/llm-providers.md

@@ -22,16 +22,26 @@ The LLM chat feature supports multiple “provider” presets, all proxied serve
 - For legacy single-model users, provider defaults to `openai_style`.
 
 ### Example provider configs (common free/trial tiers)
-- **Zhipu GLM 4.5 Flash** (OpenAI-compatible):
+- **Zhipu (Z.AI) GLM (OpenAI-compatible)**:
   - Provider: `openai_style`
   - Base URL: `https://open.bigmodel.cn/api/paas/v4`
-  - Model: `glm-4.5-flash`
+  - Endpoint note: v4 commonly supports `POST /chat/completions` (legacy `/completions` may 404)
+  - Model: e.g. `glm-4.5-flash` (availability/quotas vary)
   - Auth: `Authorization: Bearer <your-api-key>`
-- **Google Gemini 1.5 Flash** (trial tier if available):
+
+- **Google Gemini (Gemini API / AI Studio)**:
   - Provider: `gemini`
-  - Base URL: `https://generativelanguage.googleapis.com`
-  - Model: `gemini-1.5-flash`
+  - Base URL: `https://generativelanguage.googleapis.com/v1beta`
+  - Model: e.g. `gemini-2.5-flash` (check current listings)
   - Auth: `x-goog-api-key: <your-api-key>`
+  - Usage note: rate limits depend on project tier; see AI Studio for RPM/TPM/RPD.
+
+- **Gemini via OpenAI-compatible mode**:
+  - Provider: `openai_style`
+  - Base URL: `https://generativelanguage.googleapis.com/v1beta/openai/`
+  - Model: e.g. `gemini-2.5-flash`
+  - Auth: `Authorization: Bearer <your-api-key>` (as used by OpenAI SDK semantics)
+
 - **OpenAI-compatible community gateways** (if you have a key):
   - Provider: `openai_style`
   - Base URL: gateway base (e.g., `https://api.openai.com` or your proxy)