mirror of
https://github.com/openai/codex.git
synced 2026-09-05 15:18:41 +00:00
- Make `ContextManager::for_prompt` modality-aware and strip input_image content when the active model is text-only. - Added a test for multi-model -> text-only model switch