FULL STORY

xAI Launches Grok Imagine 2.0 with Advanced Editing

xAI launched Grok Imagine 2.0, topping global image generation rankings with precise editing features. The model was subsequently integrated into Vercel and officially opened to developers via API.

2026-08-08 ~ 2026-08-13 · 5 episodes · 62 posts

Episode 1 · xAI Releases Imagine Image 2.0, Ranking Second Globally in Image Generation and Editing (2026-08-08, 32 posts)

xAI officially released its new image generation model Imagine Image 2.0 (also known as Grok Image 2.0) on August 8, available as a new Quality Mode across Grok's iOS, Android, and web platforms. According to the LMSYS Chatbot Arena leaderboard, the model ranks second globally in both text-to-image (1320 points) and image editing (1439 points), behind OpenAI's GPT Image 2.0. The new model focuses on practical workflows, with significant improvements in text rendering, local editing, and multi-reference generation, sparking enthusiastic community feedback.

Confirmed

  • Leaderboard performance: According to the Arena leaderboard, Imagine Image 2.0 (Low) ranks second globally in both text-to-image and image editing, behind OpenAI's GPT Image 2.0.
  • Full platform rollout: The model is available as a new Quality Mode for users, covering Grok's iOS, Android, and web clients.
  • Core feature updates: Supports precise local editing (e.g., hover selection, Magic wand local repaint), multi-reference image editing, background replacement, and multi-aspect cropping.
  • Text rendering capability: Multiple creators' tests confirm the model achieves extremely high accuracy in rendering background text, complex layouts, and very small English text, nearly overcoming the previous pain point of blurry text in AI-generated images.

Why it matters

  • Reshaping practical workflows: Users @tetsuoai and @chaitu note that the model's local repaint and precise cutout capabilities reach a "god-level" standard, directly applicable to real commercial scenarios like group photo face swaps, meme creation, and turning e-commerce product photos into photography-style shots.
  • Enhanced interaction experience: @XFreeze mentions that users can simply hover over a specific area to automatically identify and modify it, a WYSIWYG interaction that greatly lowers the barrier to AI image editing.
  • Market competitiveness: Users like @ns123abc report that its generation experience can fully replace ChatGPT's image generation in most daily scenarios, marking xAI's rapid catch-up and solidification of its leading position in the multimodal domain.

12 more related posts →

Episode 2 · Grok Imagine Launches Precise Edit and Layer Export (2026-08-08, 9 posts)

xAI's image generation tool Grok Imagine has introduced Precise Edit and layer editing features. Users can hover over specific areas of an image to instantly redraw them while preserving other elements, and can even separate layers and download transparent PNG files. Elon Musk personally announced the update on X, and testers have expressed amazement at its precise control.

Confirmed

  • Feature launch: Grok Imagine officially introduces Precise Edit, allowing hover-based local modifications, addressing the pain point of AI editing often ruining the overall image.
  • Layer editing: The new feature supports automatic segmentation of generated image layers; users can download each independent layer as transparent PNG or edit specific layers, useful for design work requiring extraction of individual elements.
  • Executive endorsement: Musk personally retweeted and recommended the feature on X, stating that its image editing capabilities have been greatly improved.

Why it matters

  • Improved interaction efficiency: The feature makes image editing more efficient, allowing fine modifications without regenerating the entire image, greatly enhancing user experience and editing speed. User @XFreeze expressed amazement at its control; designers find the Segments panel greatly improves controllability of AI image generation; user @yuntatsai praised its multimodal grounding ability as outstanding, accurately understanding and editing image content.

Episode 3 · Vercel AI Gateway Exclusively Launches Grok Imagine 2.0 (2026-08-09, 2 posts)

Vercel has exclusively launched the preview of xAI's Grok Imagine Image 2.0 on its AI Gateway, allowing developers to access it via the npx ai-cli tool. The model currently ranks second on the image generation leaderboard.

Episode 4 · Grok Imagine 2.0 Launches, Precision Editing Ranks Second Globally (2026-08-09, 12 posts)

xAI officially released image generation model Grok Imagine 2.0 (also called Grok Image 2.0), drawing attention with native precision local editing. It quickly rose to second place globally on LMSYS text-to-image and editing leaderboards, behind GPT-image-2. Multiple testers note it has evolved from a pure generator to a practical creative tool, showing potential in UI design, movie posters, and other commercial fields.

Confirmed

  • Leaderboard: Grok Imagine 2.0 ranks second worldwide on LMSYS, behind GPT-image-2 (@eyishazyer, @SimplyAnnisa, @APPSO).
  • Precise layering and local editing: After uploading an image, the model auto-segments (e.g., face, gestures, background) and offers a magic wand for surgical edits on pointed regions. Users can download transparent-background subjects or remove backgrounds with one click (@minchoi, @markk, @APPSO).
  • Comprehensive creation: Supports combining multiple reference images, rendering clear text, and adjusting image size (@minchoi).
  • Multimodal expansion: Supports converting generated images directly to video within the same platform, with high realism (@XFreeze).
  • Official positioning: Grok officially announced the model as a major upgrade aiming to set a new standard in AI image generation (@dreamwieber).

Why it matters

Grok Imagine 2.0 breaks the limitation of generating only whole images, natively integrating advanced segmentation and interactive local editing. Bloggers @minchoi and @markk note high practical value in UI/UX hero design, movie poster generation, and hyper-realistic screenshots. Leaker @nimaowji also confirms its excellent performance, marking xAI's significant step in multimodal creative tools.

Episode 5 · Grok Imagine 2.0 Launches with Precise Editing, Multi-Image Fusion, and Low-Cost API (2026-08-11, 7 posts)

xAI's image generation tool Grok Imagine has been upgraded to version 2.0, with the API now available. The new version breaks through basic text-to-image limitations, significantly enhancing image editing capabilities and evolving into a versatile creative tool that can almost replace traditional Photoshop workflows. The API pricing is aggressive, starting at $0.04 per 1K resolution image, and developers can try it via the official Playground.

Confirmed

  • The new model supports fusing up to 5 reference images for creation, with both text-to-image and image-to-image capabilities.
  • It enables highly precise local image modifications via a single prompt, one-click background removal, and size adjustment.
  • It features clear text rendering and can generate images with cinematic quality and advanced UI elements.
  • The API handles complex scenarios such as infographics, ad creatives, game assets, and UI/UX design.

Unconfirmed

  • Specific technical details (e.g., parameter count, training data) have not been disclosed.
  • The official call for community feedback suggests future improvements, but specific directions are unclear.

Why it matters

  • Grok Imagine 2.0 expands AI image generation from simple text-to-image to fine-grained editing workflows, greatly enhancing practicality. This marks the potential for AI image tools to gradually replace traditional professional design software, and with aggressive API pricing, it can meet more complex commercial creative and development needs.