Version v0.7.7

Review generated work in a new workspace

You can now review generated images in a dedicated workspace instead of digging through chat history. Four external assistants can also place their results on the canvas, with clearer consent and better tracking of which model made each reply. Chat and canvas interactions are more reliable, so runs are less likely to get stuck or reopen by mistake.

Release highlights

What’s new

  1. Review work in one dedicated space

    Creative Review gives you a focused place to inspect generated results at scale. You get an all-items gallery, a minimal preview, and inline chat context so you can move through work faster.

  2. Use more assistants on the canvas

    ChatGPT Codex, Grok, Claude, and Gemini now work alongside the Artyx Agent. Each one shows only the capabilities it can actually use, and you can see which agent and model produced a reply even after reopening the app.

  3. Derive variations more predictably

    Quick Derive now follows the variation count you choose, so the number of images you get is easier to control. Prompts are also cleaned up before sending, which helps keep derivations focused on the result you want.

  4. Chat controls are easier to reach

    Agent, model, and connector options now sit together in one tab strip, with model choices sliding out from the right. The chat dock also responds better when you resize it, so the layout stays usable in tighter spaces.

  5. Images and 3D items attach more reliably

    Connector-placed images now behave like normal canvas items right away, so you can click, chat with, and open them without waiting for a reload. Attached 3D objects also accept more file types during analysis, which makes it easier to keep a run moving.

In the app

Where to find it

  1. 01

    Open the Creative Review workspace.

    Use the ALL gallery to scan generated work, then open a minimal preview when you want to focus on one result.

  2. 02

    Open chat and choose an assistant from the tab strip.

    You can switch between agent, model, and connector controls in one place.

  3. 03

    Start a derive action from the unified menu.

    Pick your variation count in the composer, then send the prompt.

  4. 04

    Place a generated image or 3D object on the canvas.

    You can now work with it right away, including chat follow-ups and normal image interactions.