Tyler Zhang eb2f432e16 Fix GPT-5 models to respect reasoning_effort config setting ... Previously, GPT-5 models had reasoning_effort hardcoded to 'minimal', which caused two issues: 1. 'minimal' is not supported by some models (e.g., gpt-5.1-codex) 2. User's config.reasoning_effort setting was completely ignored This fix: - Reads and respects user's reasoning_effort config value - Uses valid defaults: 'none' for non-thinking models, 'low' for thinking - Adds logging to show which value is being used Fixes #2120		2025-12-07 11:51:13 +08:00
..
base_ai_handler.py	Format files by pre-commit run -a	2024-10-30 10:00:36 +09:00
langchain_ai_handler.py	change type check and remove useless sync	2025-05-26 14:52:45 +09:00
litellm_ai_handler.py	Fix GPT-5 models to respect reasoning_effort config setting	2025-12-07 11:51:13 +08:00
litellm_helpers.py	refactor(ai_handler): remove model parameter from _get_completion and handle it within the method	2025-07-13 21:29:53 +03:00
openai_ai_handler.py	add img_path and _create_chat_async	2025-05-25 15:34:50 +09:00