Increase or adapt the file preview size limit for high-performance machines
What feature would you like to see?
Codex Desktop currently blocks file previews above a fixed 10.0 MB limit. I would like the app to increase this preview limit, or make it configurable/adaptive, so users can preview larger files directly inside Codex.
Current behavior
When opening a larger local file in Codex, the preview pane shows:
File is too large to preview
62.3 MB exceeds the 10.0 MB preview limit
This means I cannot directly inspect some larger ZIPs, PDFs, reports, or other project files inside Codex.
Why this matters
The current 10 MB cap is too restrictive for many normal workflows. Users often need to review larger files while working with Codex, and having to leave the app or open the file externally makes the workflow less convenient.
This is especially inconvenient when the file is part of a Codex task and I want to quickly preview/review it in context.
Suggested behavior
Please consider one or more of the following options:
- Increase the default preview limit.
- Add a user-configurable preview size limit in Settings.
- Add a per-file Preview anyway / Open anyway option with a performance warning.
- Automatically adjust the limit based on the user's device capabilities.
For example, on a high-performance machine with a lot of memory, the app could safely allow a much larger preview limit, such as 100 MB or more. On lower-end machines, Codex could keep a smaller default limit to avoid performance issues.
Expected behavior
Users with capable hardware should be able to preview larger files directly in Codex, or at least opt in to doing so after seeing a warning.
Environment example
- Codex Desktop
26.616.81150— Released Jun 24, 2026 - macOS Tahoe
26.5.1 - MacBook Pro 16-inch
- Apple M5 Max
- 128 GB memory
- Example blocked file size:
62.3 MB - Current displayed limit:
10.0 MB
Related context
This is related to #26805, but this issue specifically suggests an adaptive/hardware-aware preview limit so high-performance machines can use a much larger cap without forcing the same limit on all users.