> ## 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.

# April 28 – May 4, 2026

> Weekly changelog covering v1.1.0: 8-model comparisons, stacked session workflow, archived chat tabs, voice dictation, reasoning effort control, and the new image edit toolbar.

<Update label="April 28 – May 4, 2026" description="v1.1.0" tags={["Feature", "Improvement", "Fix"]}>
  ## New features

  **Compare up to 8 models per generation** — The per-generation model cap doubled from 4 to 8, so you can line up more side-by-side comparisons in a single run. This applies uniformly to image and video, and the landing page picker now supports multi-select for video too. See [multi-model generation](/image-generation/multi-model).

  **Stacked session workflow by default** — New sessions now open in the stacked layout, putting your generation surface, gallery, and chat in a single coherent view. Learn more about [working in sessions](/sessions/introduction).

  **Archived chat tabs in every session** — You can keep parallel chat explorations alongside your work. Each session now supports multiple chat tabs that auto-name themselves from your first message, so threads stay organized as you branch ideas. See [sessions](/sessions/introduction) for details.

  **Voice dictation in chat** — Speak your prompts directly into the chat input. The new dictation flow transcribes on the fly so you can ideate without typing.

  **Reasoning effort control for chat** — A new pill on the chat composer lets you set reasoning effort to Off, Low, Medium, or High. Higher levels give you deeper assistant reasoning when you need it; lower levels keep things snappy. Your choice is remembered per browser. See the [prompt assistant](/sessions/prompt-assistant).

  **Image edit toolbar (Phase 1)** — A new in-canvas toolbar adds crop, split-grid, and download actions directly to your generated images, so quick edits no longer need a round trip to another tool.

  **Free trial support** — New users can start a free trial to try premium features before picking a plan. See [plans](/billing/plans) for what's included.

  **Cross-device preferences** — Generation preferences (model, aspect ratio, video defaults, and more) now sync to your account, so your setup follows you between devices. See [settings](/image-generation/settings).

  **In-app notifications** — A new notification broadcast system delivers product updates and important account messages directly inside the app.

  ## Updates

  **Inline outpaint controls** — Outpaint now lives on the canvas surface with a slider, directional arrows, prompt input, and generate/cancel controls — replacing the old drag-handle popover for a more direct editing experience.

  **New Mux video player** — Video playback was rewritten on the official Mux player, giving cleaner autoplay, posters, and interaction handling, plus fewer console errors. See the [video generation guide](/video-generation/introduction).

  **Animated preview length preference** — A new `gifMaxFrames` preference lets you limit the frame budget for new animated video thumbnails. Manage it from your generation settings.

  **Image-first gallery** — The persistent model badge on each card was moved into a hover toolbar, paired with the provider icon. The masonry grid was also tightened so artwork leads, not labels.

  **Auto-named chat tabs** — Chat tabs now derive a 3-word title from your first message, making it faster to scan parallel threads.

  **Streamlined generation toolbar** — Model selector ordering was prioritized for the most-used controls, and the input scroll area was reworked so long prompts no longer push the toolbar around.

  ## Bug fixes

  * **Toolbar text bleed-through** — Long prompts no longer leak visually behind the model selector and settings pills.
  * **Outpainted images in history** — Outpainted results now render correctly in the generation history view.
  * **Session intro prompt flicker** — Fixed a race where the intro prompt could be wiped mid-stream while chat tabs were still resolving.
  * **Assist image plan ratios** — Malformed aspect ratios from the assistant are now handled gracefully instead of failing the plan.
  * **Prompt history & local sessions** — Prompt history linkage and local auth sessions are now preserved correctly across sign-ins.
</Update>
