Gemma4 and dep upgrades

.devcontainer/ollama/devcontainer.json

@@ -9,10 +9,10 @@
     "ghcr.io/devcontainers/features/docker-in-docker:2": {},
     "ghcr.io/azure/azure-dev/azd:latest": {},
     "ghcr.io/prulloac/devcontainer-features/ollama:1": {
-      "pull": "qwen3.5:9b"
+      "pull": "gemma4:e2b"
     }
   },
-  "postCreateCommand": "uv sync && cp .env.sample.ollama .env",
+  "postCreateCommand": "uv sync && uv run prek install && cp .env.sample.ollama .env",
   "forwardPorts": [6277, 6274],
   "portsAttributes": {
     "6277": {

.env-sample

@@ -7,7 +7,7 @@ AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com/
 AZURE_OPENAI_CHAT_DEPLOYMENT=your-deployment-name
 
 # Ollama Configuration
-OLLAMA_MODEL=qwen3.5:9b
+OLLAMA_MODEL=gemma4:e2b
 OLLAMA_ENDPOINT=http://localhost:11434/v1
 OLLAMA_API_KEY=no-key-needed
 

.env.sample.ollama

@@ -2,7 +2,7 @@
 API_HOST=ollama
 
 # Ollama Configuration
-OLLAMA_MODEL=qwen3.5:9b
+OLLAMA_MODEL=gemma4:e2b
 OLLAMA_ENDPOINT=http://localhost:11434/v1
 OLLAMA_API_KEY=no-key-needed
 

README.md

@@ -36,7 +36,7 @@ You can run this project virtually by using GitHub Codespaces. Click one of the
 
 [![Open in GitHub Codespaces](https://github.com/codespaces/badge.svg)](https://codespaces.new/Azure-Samples/python-mcp-demos?devcontainer_path=.devcontainer/ollama/devcontainer.json)
 
-The Ollama Codespace pre-installs Ollama and pulls the `qwen3.5:9b` model, and copies `.env.sample.ollama` as your `.env` file. Note that the 64GB memory requirement will consume your Codespace quota faster.
+The Ollama Codespace pre-installs Ollama and pulls the `gemma4:e2b` model, and copies `.env.sample.ollama` as your `.env` file. Note that the 64GB memory requirement will consume your Codespace quota faster.
 
 Once the Codespace is open, open a terminal window and continue with the deployment steps.
 

agents/agentframework_http.py

@@ -39,7 +39,7 @@
     client = OpenAIResponsesClient(
         base_url=os.environ.get("OLLAMA_ENDPOINT", "http://localhost:11434/v1"),
         api_key=os.getenv("OLLAMA_API_KEY", "no-key-needed"),
-        model_id=os.environ.get("OLLAMA_MODEL", "qwen3.5:9b"),
+        model_id=os.environ.get("OLLAMA_MODEL", "gemma4:e2b"),
     )
 elif API_HOST == "openai":
     client = OpenAIResponsesClient(

agents/agentframework_learn.py

@@ -32,7 +32,7 @@
     client = OpenAIResponsesClient(
         base_url=os.environ.get("OLLAMA_ENDPOINT", "http://localhost:11434/v1"),
         api_key=os.getenv("OLLAMA_API_KEY", "no-key-needed"),
-        model_id=os.environ.get("OLLAMA_MODEL", "qwen3.5:9b"),
+        model_id=os.environ.get("OLLAMA_MODEL", "gemma4:e2b"),
     )
 elif API_HOST == "openai":
     client = OpenAIResponsesClient(
@@ -58,7 +58,7 @@ async def http_mcp_example() -> None:
                 tools=[mcp_server],
             ) as agent,
         ):
-            query = "How to create an Azure storage account using az cli?"
+            query = "What are the available hosting options for a Python web app on Azure? Compare them briefly."
             result = await agent.run(query)
             print(result.text)
 

agents/langchainv1_github.py

@@ -36,7 +36,7 @@
     )
 elif API_HOST == "ollama":
     model = ChatOpenAI(
-        model=os.environ.get("OLLAMA_MODEL", "qwen3.5:9b"),
+        model=os.environ.get("OLLAMA_MODEL", "gemma4:e2b"),
         base_url=os.environ.get("OLLAMA_ENDPOINT", "http://localhost:11434/v1"),
         api_key=SecretStr(os.getenv("OLLAMA_API_KEY", "no-key-needed")),
         use_responses_api=True,
@@ -92,7 +92,7 @@ async def main():
     agent = create_agent(
         model,
         tools=filtered_tools,
-        prompt="You help users research GitHub repositories. Search and analyze information.",
+        system_prompt="You help users research GitHub repositories. Search and analyze information.",
     )
 
     query = "Make a list of last 5 issues from the 'PrefectHQ/FastMCP' repository that discuss auth."

agents/langchainv1_http.py

@@ -38,7 +38,7 @@
     )
 elif API_HOST == "ollama":
     base_model = ChatOpenAI(
-        model=os.environ.get("OLLAMA_MODEL", "qwen3.5:9b"),
+        model=os.environ.get("OLLAMA_MODEL", "gemma4:e2b"),
         base_url=os.environ.get("OLLAMA_ENDPOINT", "http://localhost:11434/v1"),
         api_key=SecretStr(os.getenv("OLLAMA_API_KEY", "no-key-needed")),
         use_responses_api=True,

pyproject.toml

@@ -7,14 +7,14 @@ requires-python = "==3.13.*"
 dependencies = [
     "fastmcp>=3.0.0",
     "debugpy>=1.8.0",
-    "langchain-core>=0.3.0",
+    "langchain-core>=1.2.26",
     "mcp>=1.3.0",
     "azure-identity>=1.25.1",
     "msgraph-sdk>=1.0.0",
     "dotenv-azd>=0.1.0",
-    "langchain>=1.0.0",
-    "langchain-openai>=1.0.1",
-    "langchain-mcp-adapters>=0.1.11",
+    "langchain>=1.2.15",
+    "langchain-openai>=1.1.12",
+    "langchain-mcp-adapters>=0.2.2",
     "azure-ai-agents>=1.1.0",
     "agent-framework-core==1.0.0rc5",
     "azure-cosmos>=4.9.0",

spanish/README.md

@@ -35,7 +35,7 @@ Puedes ejecutar este proyecto de forma virtual usando GitHub Codespaces. Haz cli
 
 [![Open in GitHub Codespaces](https://github.com/codespaces/badge.svg)](https://codespaces.new/Azure-Samples/python-mcp-demos?devcontainer_path=.devcontainer/ollama/devcontainer.json)
 
-El Codespace de Ollama pre-instala Ollama y descarga el modelo `qwen3.5:9b`, y copia `.env.sample.ollama` como tu archivo `.env`. Ten en cuenta que el requisito de 64GB de memoria consumirá tu cuota de Codespace más rápido.
+El Codespace de Ollama pre-instala Ollama y descarga el modelo `gemma4:e2b`, y copia `.env.sample.ollama` como tu archivo `.env`. Ten en cuenta que el requisito de 64GB de memoria consumirá tu cuota de Codespace más rápido.
 
 Una vez abierto el Codespace, abre una terminal y continúa con los pasos de despliegue.