Upgrade to gemini-3-flash-preview (#1)

* CLAUDE.md intro

* Fix broken documentation rendering

* Upgrade to gemini-3-flash-preview

Author: krasserm

CLAUDE.md

@@ -1,3 +1,7 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
 ## Overview
 
 The `group_sense` package is a library for detecting patterns in group chat message streams and transforming them into self-contained queries for downstream AI systems. This enables existing single-user AI agents to participate in group conversations based on configurable criteria, without requiring training on multi-party conversations.

examples/chat/assistant.py

@@ -26,7 +26,7 @@ def __init__(self):
         self._history: list[ModelMessage] = []
         self._agent = Agent(
             system_prompt=SYSTEM_PROMPT,
-            model="gemini-2.5-flash",
+            model="google-gla:gemini-2.5-flash",
             model_settings=GoogleModelSettings(
                 google_thinking_config={
                     "thinking_budget": 0,

group_sense/reasoner/default.py

@@ -47,7 +47,7 @@ def __init__(
         Args:
             system_prompt: System prompt that defines the reasoner's behavior and
                 decision-making criteria. Should not contain an {owner} placeholder.
-            model: Optional AI model to use. Defaults to "gemini-2.5-flash".
+            model: Optional AI model to use. Defaults to "google-gla:gemini-3-flash-preview".
                 Can be a model name string or a pydantic-ai Model instance.
             model_settings: Optional model-specific settings. Defaults to
                 GoogleModelSettings with thinking enabled.
@@ -58,11 +58,11 @@ def __init__(
         self._agent = Agent(
             system_prompt=system_prompt,
             output_type=NativeOutput(Response),
-            model=model or "gemini-2.5-flash",
+            model=model or "google-gla:gemini-3-flash-preview",
             model_settings=model_settings
             or GoogleModelSettings(
                 google_thinking_config={
-                    "thinking_budget": -1,
+                    "thinking_level": "high",
                     "include_thoughts": True,
                 }
             ),
@@ -99,7 +99,11 @@ async def process(self, updates: list[Message]) -> Response:
         result = await self._agent.run(reasoner_prompt, message_history=self._history)
         self._history = result.all_messages()
         self._processed += len(updates)
-        return result.output
+
+        response = result.output
+        if response.receiver == "":
+            response.receiver = None
+        return response
 
     def get_serialized(self) -> dict[str, Any]:
         """Serialize the reasoner's state for persistence.

mkdocs.yml

@@ -5,7 +5,7 @@ site_url: https://gradion-ai.github.io/group-sense/
 repo_name: gradion-ai/group-sense
 repo_url: https://github.com/gradion-ai/group-sense
 
-copyright: Copyright © 2025 Gradion AI
+copyright: Copyright © 2025-2026 Gradion AI
 
 theme:
   name: material
@@ -87,10 +87,7 @@ plugins:
           - api/reasoner.md: Reasoner interfaces and implementations
 
 markdown_extensions:
-  - pymdownx.highlight:
-      anchor_linenums: true
-      line_spans: __span
-      pygments_lang_class: true
+  - pymdownx.highlight
   - pymdownx.inlinehilite
   - pymdownx.snippets:
       dedent_subsections: true

pyproject.toml

@@ -9,7 +9,8 @@ requires-python = ">=3.11,<3.15"
 readme = "README.md"
 license = "Apache-2.0"
 dependencies = [
-    "pydantic-ai>=1.3.0",
+    "pydantic-ai>=1.36.0",
+    "google-genai>=1.56.0"
 ]
 
 [tool.uv]