fix: unified_search + ai_proactive get API key from DB, fix model names for Ollama Cloud

- query_understanding.py: get API key/base_url/provider_type from ai_providers DB
- embedding.py: get API key from DB, pass db+tenant_id through call chain
- routes.py: pass db+tenant_id to llm_analyze_query and llm_aggregate_results
- search_engine.py: pass db+tenant_id to generate_embedding
- unified_search/jobs.py: pass db+tenant_id to generate_embedding
- Fix all default model names: ollama/deepseek-v4 -> ollama/deepseek-v4-flash
- Ollama Cloud has no embedding endpoint; embedding calls fail gracefully

app/plugins/builtins/ai_proactive/jobs.py

@@ -186,7 +186,7 @@ async def deep_analysis(
             extended_context["similar"] = {}
 
         # Generate extended suggestion with deep analysis prompt
-        model = settings.model or "ollama/deepseek-v4"
+        model = settings.model or "ollama/deepseek-v4-flash"
 
         # Get API key from DB (like ai_assistant does) or fall back to env
         from app.plugins.builtins.ai_proactive.services import _get_llm_api_key
@@ -195,7 +195,7 @@ async def deep_analysis(
             api_key = os.environ.get('API_KEY_OLLAMA_CLOUD', '') or None
 
         # Build model string with provider prefix
-        model = settings.model or "ollama/deepseek-v4"
+        model = settings.model or "ollama/deepseek-v4-flash"
         if provider_type:
             model_parts = model.split("/", 1)
             model = f"{provider_type}/{model_parts[-1]}"

app/plugins/builtins/ai_proactive/models.py

@@ -108,4 +108,4 @@ class ProactiveSettings(Base, TenantMixin):
     rate_limit_seconds: Mapped[int] = mapped_column(
         Integer, nullable=False, default=10
     )
-    model: Mapped[str] = mapped_column(String(100), nullable=False, default="ollama/deepseek-v4")
+    model: Mapped[str] = mapped_column(String(100), nullable=False, default="ollama/deepseek-v4-flash")

app/plugins/builtins/ai_proactive/schemas.py

@@ -74,7 +74,7 @@ class SettingsResponse(BaseModel):
     rate_limit_seconds: int
     model: str
     available_models: list[str] = Field(default_factory=lambda: [
-        'ollama/deepseek-v4',
+        'ollama/deepseek-v4-flash',
         'ollama/deepseek-v4-pro',
         'ollama/llama3.2',
         'ollama/gpt-4o-mini',

app/plugins/builtins/ai_proactive/services.py

@@ -113,7 +113,7 @@ async def get_user_settings(
             suggestion_categories=["mail", "tasks", "contacts", "companies", "insights"],
             confidence_threshold=0.5,
             rate_limit_seconds=10,
-            model="ollama/deepseek-v4",
+            model="ollama/deepseek-v4-flash",
         )
         db.add(settings)
         await db.flush()
@@ -395,7 +395,7 @@ async def generate_suggestion(
     Returns dict with suggestion_type, title, content, confidence, actions
     or None on failure.
     """
-    model = settings.model or "ollama/deepseek-v4"
+    model = settings.model or "ollama/deepseek-v4-flash"
 
     # Get API key from DB (like ai_assistant does) or fall back to env
     api_key = None
@@ -407,7 +407,7 @@ async def generate_suggestion(
         api_key = os.environ.get('API_KEY_OLLAMA_CLOUD', '') or None
 
     # Build model string with provider prefix (like ai_assistant build_litellm_params)
-    model = settings.model or "ollama/deepseek-v4"
+    model = settings.model or "ollama/deepseek-v4-flash"
     if provider_type:
         model_parts = model.split("/", 1)
         model = f"{provider_type}/{model_parts[-1]}"