01
Better prompt following
Give the model a structured brief with subjects, text, composition, and visual constraints instead of relying on vague style keywords.
Ranked #2 on Arena at launch · Preliminary snapshot
Imagine Image 2.0, in a workspace built for real image work
Imagine2 brings generation and editing into a focused browser studio. Move from a written brief to deliberate image variations, then refine the strongest result.
Imagine2 creative engine
Text to Image
Describe the subject, composition, light, style, and final use, then choose a 1K or 2K result.
Prompt input0/2000
Output darkroom
ReadyYour contact sheet starts here.
Results stay private, expire after 24 hours, and can be saved for 30 days.
Imagine2 is an independent product and is not affiliated with or endorsed by xAI.
Visual directions
Imagine Image 2.0 visual directions
Selected Imagine Image 2.0 image and prompt directions from the Gallery, spanning portraits, anime, illustration, and game-inspired scenes.
Built for more than a good-looking first draft
Imagine Image 2.0 moves the Grok image workflow toward practical design work: clearer direction, stronger layouts, targeted edits, and continuity across references.
02
Text and layout
Develop visual concepts that depend on readable text, hierarchy, and multi-part editorial layouts.
03
Precise editing
Refine an existing image instead of starting over every time one detail is wrong.
04
Reference consistency
Use source images to guide subjects, products, style, and visual continuity across iterations.
Three starting points
One workspace, three ways to start
Image generation is rarely a single prompt followed by a finished asset. Imagine Image 2.0 fits a loop of creation, editing, and reference composition. Imagine2 keeps those steps close together so the strongest result can become the source for the next deliberate change.
01
Text to image
Use Imagine Image 2.0 from a description. Choose the intended format, write a visual brief, and generate one, two, or four directions without uploading a reference.
Open Image Generator02
Image editing
Edit with Imagine Image 2.0 by uploading an existing image and describing what should change. State what must remain stable so the edit has a clear visual contract.
Open Image Editor03
Multi-image references
Guide Imagine Image 2.0 with a subject, style, product, or environment when one source cannot communicate the complete direction. Use two or three references with clearly assigned roles.
Open Multi-Image EditorArena · August 7, 2026
A major jump in the Arena image rankings
At launch, Imagine Image 2.0 ranked #2 on Arena's overall text-to-image leaderboard and #2 on the single-image-edit leaderboard. The scores were marked preliminary, so this is a dated launch snapshot rather than a permanent quality claim.
View the current Arena leaderboard- Text to Image
- #2 at launch
- Image Editing
- #2 at launch
- Source
- Arena · Aug 7, 2026
- Status
- Preliminary snapshot
Model clarity
Keep Imagine Image 2.0 results traceable
Imagine Image 2.0 sets a useful direction for practical image work: stronger instructions, readable layouts, deliberate editing, and better continuity between references. Imagine2 turns that direction into a browser workspace while keeping the selected model visible. A result made with Grok Imagine is labeled as Grok Imagine rather than being presented as output from a different model.
That distinction matters when an AI image generator is part of a real creative process. Prompts, model choice, frame, resolution, and references all affect what can be reproduced later. Imagine2 keeps those decisions close to the image editor, so a useful draft can move into a targeted edit without losing its working context. It is an AI image generator for controlled iteration, not an anonymous feed of unrelated results.
- Clear attribution
- The selected model is shown in the workspace and on completed results.
- Private by default
- References and outputs stay private unless you choose to publish them elsewhere.
- Review before reuse
- Text, geometry, identity, product details, and factual claims still require human review.
A controlled loop
Use Imagine Image 2.0 in three steps
- 01
Describe the result
Write the subject, environment, composition, lighting, and any text that should appear. In an edit, also name the parts that must remain unchanged.
- 02
Choose your format
Set an aspect ratio for the final destination, choose 1K for exploration or 2K for a more resolved output, and attach only useful references.
- 03
Generate, then refine
Review the result against the brief. Keep the strongest direction and continue with an edit instead of rewriting the entire idea from scratch.
Prompt directions worth developing
An Imagine Image 2.0 prompt is more than a style label. It names the subject, environment, composition, light, visual treatment, text or layout, and the intended output. The Gallery pairs reviewed visual directions with their full prompts, settings, and remix paths so each example can be inspected rather than merely liked.
Browse gallery examplesProduct photography
Advertising
Typography
Portrait
Anime
Illustration
Game assets
UI and icons
Imagine Image 2.0 FAQ
What Imagine2 can create, how editing and references work, and what to review before publishing.
Start creating
Turn a clear direction into a working image.
Begin in the public workspace, review the estimated credit cost, and sign in with Google only when you are ready to generate. New accounts receive ten credits with no card required.
About Imagine2. Imagine2 is an independent AI image creation tool. We build browser-based workflows on top of third-party model APIs and identify the model used for each verified generation. Imagine2 is not affiliated with, endorsed by, or sponsored by xAI.
