Codex 0.168.0 adds a side-by-side workspace pane, audio recording attachments for audio-capable models, and per-session model provider handling. These change how engineers run and steer long coding sessions.
Key takeaways · AI-distilled
Codex 0.168.0 deprecates the `features.multi_AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition →` flag: Codex now warns when it is set explicitly and points users to `agents.enabled` to turn off multi-agentUsing several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.Full definition → tools.
Steering messages sent during the final transcript save, or after switching back to a long-running chat, now reach the running turn, per the bug-fix notes.
Fuzzy file search now gives filename matches more weight, so results where only a parent directory matches the query are less likely to crowd out the file you named.
The full changelog adds per-turn prompt cache TTL options (#59480) and a warning when project instructions exceed the configured byte budget (#59445).
Animated GIF generation and editing, with optional frame count and timing controls, ships behind the `features.image_generation_animation` flag.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
multi-agent — Using several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.
