Add remote OpenAI-compatible OpenChat example (DaoXE)

docs/components/model_catalog.md

@@ -195,6 +195,36 @@ ModelCatalog().register_open_chat_model("my_open_chat_model2",
                                         model_type="chat")
 ```
 
+# Add a remote OpenAI-compatible endpoint (DaoXE)
+
+The same `register_open_chat_model` path works for **remote** OpenAI-compatible Chat Completions APIs, not only local LM Studio.  
+[DaoXE](https://daoxe.com) is a multi-model, multi-protocol API gateway: OpenAI Chat Completions / Responses **and** Anthropic Messages (Claude protocol), plus other catalog endpoints. This snippet uses the OpenAI-compatible surface that `OpenChatModel` already supports.
+
+```python
+import os
+from llmware.models import ModelCatalog
+from llmware.prompts import Prompt
+
+# Create a key at https://daoxe.com — model IDs come from your live catalog / account.
+# DaoXE does not serve users in mainland China; availability can vary by region and account.
+api_key = os.environ["DAOXE_API_KEY"]
+model_id = os.environ["DAOXE_MODEL"]          # real model ID from your DaoXE catalog
+api_base = os.environ.get("DAOXE_BASE_URL", "https://daoxe.com/v1")
+
+# model_name is both the llmware catalog key and the remote `model` field
+ModelCatalog().register_open_chat_model(
+    model_name=model_id,
+    api_base=api_base,
+    prompt_wrapper="",
+    model_type="chat",
+)
+
+prompter = Prompt().load_model(model_id, api_key=api_key)
+response = prompter.prompt_main("What is the future of AI?")
+```
+
+Full runnable example: [`solutions/models/using-daoxe-remote.py`](https://github.com/llmware-ai/llmware/blob/main/solutions/models/using-daoxe-remote.py).
+
 
 Need help or have questions?
 ============================

solutions/models/using-daoxe-remote.py

@@ -0,0 +1,83 @@
+"""
+    This example shows how to use a remote OpenAI-compatible Chat Completions endpoint with
+    llmware's OpenChatModel path (register_open_chat_model + api_base).
+
+    DaoXE is a multi-model, multi-protocol API gateway. Besides OpenAI-compatible Chat
+    Completions / Responses, it also exposes Anthropic Messages (Claude protocol) and other
+    catalog endpoints. This example focuses on the OpenAI-compatible surface that llmware
+    already supports via OpenChatModel.
+
+    Setup:
+        1. Create an API key at https://daoxe.com (availability can vary by account / region;
+           DaoXE does not serve users in mainland China).
+        2. Export:
+
+            export DAOXE_API_KEY="your_api_key"
+            export DAOXE_MODEL="model_id_from_your_catalog"   # use an ID from your live catalog
+
+        Optional override for the base URL (default is https://daoxe.com/v1):
+
+            export DAOXE_BASE_URL="https://daoxe.com/v1"
+
+    Notes:
+        - register_open_chat_model uses model_name both as the llmware catalog key and as the
+          `model` field sent to the remote API, so set DAOXE_MODEL to a real catalog model ID.
+        - OpenChatModel reads the API key from load_model(..., api_key=...) or
+          USER_MANAGED_OPEN_CHAT_API_KEY; we pass DAOXE_API_KEY explicitly below.
+        - Do not hard-code model IDs: DaoXE model availability depends on the live catalog
+          and the caller's account.
+"""
+
+import os
+import sys
+
+from llmware.models import ModelCatalog
+from llmware.prompts import Prompt
+
+
+def main():
+    api_key = os.environ.get("DAOXE_API_KEY")
+    model_id = os.environ.get("DAOXE_MODEL")
+    api_base = os.environ.get("DAOXE_BASE_URL", "https://daoxe.com/v1")
+
+    if not api_key or not model_id:
+        print(
+            "Set DAOXE_API_KEY and DAOXE_MODEL before running this example.\n"
+            "  export DAOXE_API_KEY='your_api_key'\n"
+            "  export DAOXE_MODEL='model_id_from_your_catalog'\n"
+            "Optional:\n"
+            "  export DAOXE_BASE_URL='https://daoxe.com/v1'"
+        )
+        sys.exit(1)
+
+    # Register the remote OpenAI-compatible endpoint in the Model Catalog.
+    # prompt_wrapper="" keeps the chat messages path without extra local instruct wrappers
+    # (remote chat models typically expect normal role/content messages).
+    ModelCatalog().register_open_chat_model(
+        model_name=model_id,
+        api_base=api_base,
+        prompt_wrapper="",
+        model_type="chat",
+        display_name=f"daoxe:{model_id}",
+    )
+
+    # Confirm registration
+    card = ModelCatalog().lookup_model_card(model_id)
+    print("update: registered open-chat model card -", card)
+
+    # Load with the remote API key (OpenChatModel also accepts USER_MANAGED_OPEN_CHAT_API_KEY)
+    prompter = Prompt().load_model(model_id, api_key=api_key)
+    response = prompter.prompt_main("In one short sentence, what is retrieval-augmented generation?")
+
+    print("update: daoxe remote response -", response)
+
+    # Direct ModelCatalog path (same registration)
+    model = ModelCatalog().load_model(model_id, api_key=api_key)
+    direct = model.inference("Reply with a single word: ready")
+    print("update: daoxe direct inference -", direct)
+
+    return 0
+
+
+if __name__ == "__main__":
+    main()