Documentation Index
Fetch the complete documentation index at: https://docs.opencanva.co/llms.txt
Use this file to discover all available pages before exploring further.
New features
Smart queue overflow buffering — When your generation queue is full, new jobs are now automatically buffered on your device and retried as slots open up. Previously, hitting your queue depth limit meant your submission was rejected and you had to try again manually. Now you can keep generating without interruption — the system handles the rest.Generation toolbar status — A new status indicator in the generation toolbar shows your active, queued, and buffered generation counts at a glance. You can see exactly how many jobs are running, waiting on the server, and buffered locally, making it easy to manage busy workflows without switching views. Learn more about generation status.Updates
Smarter prompt assistant model matching — The prompt assistant now validates that recommended models match your selected media type. If you’re generating images, the assistant will only suggest image models — and vice versa for video. When a mismatch is detected, you’ll see a clear, friendly message instead of a broken settings panel.Graceful missing reference images — When a reference image has been deleted or is no longer available, it now appears dimmed with a tooltip explaining the issue. Interactive actions like selecting or removing the reference are disabled, preventing confusion. Previously, missing references could appear as broken thumbnails with no explanation.Chat history compatibility — Older conversations from earlier versions of the prompt assistant now load correctly regardless of their original format. If you had chat sessions that previously showed blank or failed to render, they should now display as expected.Bug fixes
- Prompt assistant error display — When the AI assistant encounters an issue generating a plan, you now see a concise error message with optional technical details instead of raw JSON output.
- Video polling efficiency — Long-running video generations now poll for status updates with smarter timing, reducing unnecessary network requests and improving responsiveness when jobs complete.
- Queue prompt sync — Fixed an issue where editing a queued generation’s prompt would not update the prompt shown in your session history, causing a mismatch between what you edited and what appeared after the job ran.