You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Vector stores. In openai vector-stores manage, use / to find a loaded store by name or ID. Browse local files or paste a path, then watch indexing progress. Long questions wrap, and wide terminals show matching passages beside results. Press Ctrl+G to expand the workspace or start with --full-screen. #746
Batches. Press / to filter batches by ID, status, or endpoint. In completed CSV batches, search source rows and answers, then open a match. #742, #745
Explore. Press Ctrl+G to expand command search. The picker keeps your query and selection. You can also start expanded with openai explore --full-screen. #748
Speech and images. New speech drafts use gpt-realtime-2.1-mini. The CLI's image saving preset uses gpt-image-2.6-sunburst for generation and edits. OpenAI image variation requests now show how to use image edits. #729
Responses API. The schema describes the original item ID and removal reason reported by reasoning input transformations. #749
Bug fixes
Voice Studio. Shortcuts now sit below the panel. Startup avoids stray characters in Apple Terminal. #750
Images. Fixed gaps in iTerm2 previews. After Ctrl+C, trackpad scrolling returns to terminal history. #735
Tokenizer. Restores the shell viewport and visible history after resizing in the Linux VS Code terminal. #744
Documentation
Updated file transcription examples to use gpt-transcribe for plain text and JSON. #743
Documented how to expand the vector store workspace in the Linux VS Code terminal. #747
Release checks
The release workflow now verifies the CLI's default models against OpenAI's current guidance before publishing. #739