Grok quietly launches Imagine Agent Mode: Infinite canvas replaces chat box, generating entire sets of images and videos with a single prompt

📄Full Article· Automatically extracted by trafilaturaGemini 翻譯1054 words

Multiple users have discovered that the web version of Grok is testing Imagine Agent Mode (beta), where the interactive interface has shifted from a traditional chat box to an infinite canvas, allowing users to generate entire short films, comics, or product content packages with a single prompt. Since the launch of Grok Imagine in February this year, over 1.245 billion videos have been generated within 30 days. xAI has not yet released an official announcement. (Previous coverage: Musk quietly shuts down Starlink customer service centers: Grok Voice takes over calls, 20% of calls closed directly) (Background: Altman reveals at Stripe conference: OpenAI aims to be "forever low-margin" intelligent infrastructure, with switching costs so low that "you can move just by telling an agent") Users recently discovered on X that the web version of Grok is secretly testing a new feature called Imagine Agent Mode. xAI has not yet released any official announcements, but judging from user screenshots, this is not a typical feature update. The entire interactive interface has been replaced from a traditional chat box to an "infinite canvas," where the agent can generate images and videos step-by-step. Generating a full set of assets with one sentence User screenshots show that a new "Imagine" entry has been added to the Grok sidebar. Clicking it opens an open canvas, with default workflow templates listed on the right: Create Worlds, Short Film, UGC Product Stories, and Brand Identity. By describing requirements in a single sentence, the agent generates all materials on the canvas in sequence. Specific capabilities "discovered" by users include generating and editing multiple images simultaneously, converting static images into videos and automatically stitching them together, video cropping and fading, and exporting finished products. TestingCatalog pointed out that Imagine Agent can handle complex, multi-step instructions, such as "generate a 1-minute short film," "generate a full set of comics," or "generate a UGC product content package." Another user demonstrated the complete process where the agent first batch-generates 3 sets of product images and 3 sets of model photos, then automatically assembles them into a social media content package. When Grok Imagine launched its 1.0 version in February this year, it generated over 1.245 billion videos within 30 days, utilizing the Aurora engine, supporting 720p resolution, 10-second videos, and native audio. What Agent Mode adds to this foundation is workflow orchestration capability—not just letting the model "generate better things," but allowing it to decide for itself what to generate, in what order, and how to combine them. This is the first time xAI has extended its agent architecture to the field of creative generation since the launch of Grok Build's parallel agents. Although it has not been officially released yet, it may be a regional test or a preliminary trial, and the feature may not necessarily be implemented in the final version.

Data Status✓ Full text extractedRead Original (動區 BlockTempo)

🔍Historical Similar Events· Keyword + Asset Matching6 items

2026-04-28

NVIDIA launches new open-source multimodal large model "Nemotron 3 Nano Omni"! Capable of processing video, audio, images, and text, specialized for Agent applications.

Similarity 120%關鍵字 agent同分類 zh

2026-04-28

Meituan has quietly launched its new-generation large model "LongCat-2.0-Preview"! It sets a record for the largest training computing power in China and specializes in AI Agent.

Similarity 120%關鍵字 agent同分類 zh

2026-04-27

Musk quietly kills Starlink customer service centers: Grok Voice takes over calls, 20% of calls closed directly

Similarity 120%關鍵字 grok同分類 zh

2026-04-25