[mod] ai_summary plugin: switch to the OpenAI chat completions API

Talk to the LLM server via GET /v1/models and POST /v1/chat/completions
(SSE) instead of Ollama's native API.  Any OpenAI compatible server now
works (Ollama, vLLM, llama.cpp, LM Studio, Hugging Face TGI, ...);
Ollama serves this API natively, existing setups keep working unchanged.

The Ollama specific keep_alive option is dropped, the ai_summary.grounding
setting is added as instance wide default of the grounding preference.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

searx/ai_summary.py

@@ -7,7 +7,7 @@
 # - https://github.com/searxng/searxng/issues/5284
 from __future__ import annotations
 
-__all__ = ["SettingsAISummary", "MODELS", "model_choices", "build_ollama_messages"]
+__all__ = ["SettingsAISummary", "MODELS", "model_choices", "build_chat_messages"]
 
 import msgspec
 
@@ -43,9 +43,11 @@ class SettingsAISummary(msgspec.Struct, kw_only=True, forbid_unknown_fields=True
     """
 
     base_url: str = ""
-    """Default base URL of the Ollama server (e.g. ``http://127.0.0.1:11434``).
-    Users can set their own server URL in the preferences
-    (``ai_summary_server``) unless that preference is locked."""
+    """Default base URL of the LLM server (e.g. ``http://127.0.0.1:11434``
+    for Ollama).  Any server that implements the OpenAI chat completions API
+    works (Ollama, vLLM, llama.cpp, LM Studio, Hugging Face TGI, ...).  Users
+    can set their own server URL in the preferences (``ai_summary_server``)
+    unless that preference is locked."""
 
     model: str = ""
     """Name of the default model (e.g. ``llama3.2:3b``).  If empty, the first
@@ -56,7 +58,7 @@ class SettingsAISummary(msgspec.Struct, kw_only=True, forbid_unknown_fields=True
     models: list[str] = []
     """List of model names suggested to the user in the preferences.  If empty
     and :py:obj:`SettingsAISummary.base_url` is set, the list is requested
-    once at application setup from the Ollama server (``GET /api/tags``)."""
+    once at application setup from the LLM server (``GET /v1/models``)."""
 
     grounding: bool = False
     """Default of the ``ai_summary_grounding`` user preference: ground the
@@ -64,7 +66,7 @@ class SettingsAISummary(msgspec.Struct, kw_only=True, forbid_unknown_fields=True
     preferences unless that preference is locked."""
 
     connect_timeout: float = 5.0
-    """Timeout (seconds) to establish a TCP connection to the Ollama server."""
+    """Timeout (seconds) to establish a TCP connection to the LLM server."""
 
     read_timeout: float = 30.0
     """Maximum gap (seconds) between two chunks of the token stream."""
@@ -72,10 +74,6 @@ class SettingsAISummary(msgspec.Struct, kw_only=True, forbid_unknown_fields=True
     stream_timeout: float = 120.0
     """Wall clock limit (seconds) for one completion."""
 
-    keep_alive: str = "5m"
-    """How long the model stays loaded in memory after the request (passed
-    through to Ollama's ``keep_alive`` option)."""
-
     max_context_items: int = 5
     """Maximum number of search results accepted as grounding context."""
 
@@ -99,13 +97,13 @@ def model_choices() -> list[str]:
     return list(MODELS)
 
 
-def build_ollama_messages(
+def build_chat_messages(
     cfg: SettingsAISummary,
     messages: list[dict[str, str]],
     context: list[dict[str, str]] | None = None,
 ) -> list[dict[str, str]]:
-    """Build the message list for Ollama's ``/api/chat`` from the (already
-    validated) request ``messages``, prepending a system prompt.  When
+    """Build the message list for the chat completions request from the
+    (already validated) request ``messages``, prepending a system prompt.  When
     ``context`` items are given, the grounded system prompt is used and the
     context items are serialized into its ``{context}`` placeholder."""