Tag
Elon Musk's xAI has reassigned Starlink engineer Jack Garabedian to lead the Grok training team, replacing Diego Pasini, as part of broader integration of Starlink executives into xAI management ahead of SpaceX's IPO.
xAI is increasingly acting as a datacentre REIT by renting compute capacity to Anthropic and Google, potentially recouping billions in capex within 18 months. The deals highlight the ongoing AI compute crunch and may serve strategic purposes beyond pure commerce.
Latest Grok Build v0.2.31 fixes a grep timeout issue and includes bug fixes for marketplace skills and prompt queue ordering.
User criticizes Musk for counting X's Blue subscription fees into xAI's revenue, and predicts that SpaceX's IPO may open high and then decline.
Elon Musk announces that Grok now supports worktrees, allowing parallel agents in separate workspaces.
The updated Grok model (0.5T) is less lazy, more autonomous, and more accurate; improvements are ongoing.
A UK Member of Parliament is suing Elon Musk's xAI over deepfake content, creating a landmark legal test for AI company accountability regarding synthetic media. The case could set significant precedent for how AI companies are held responsible for harmful generated content.
xAI partners with Cloudflare to make Grok models (LLMs, audio, image, video) available through Cloudflare AI Gateway, billed directly through Cloudflare with no extra authentication needed.
Elon Musk announced that the Iliad (Troy) trailer was created using the newly released Grok Imagine 1.5 text-to-video model.
xAI asks a court to force the public identification of four plaintiffs who are suing over deepfake nudes allegedly created using the Grok AI model, arguing against their anonymity.
Ethan He from xAI discusses why video agent models are the next frontier, arguing that video models derive intelligence from LLMs and that the evolution of video generation will mirror AI coding, shifting from one-shot output to multi-turn planning and execution.
A Latent Space podcast episode discusses the thesis that video models derive intelligence from LLMs, and that the next frontier is video agents. Guest Ethan He, who built Grok Imagine at xAI, shares insights on building frontier image and video systems.
A tweet praising a podcast episode where former xAI world model lead Ethan He provides deep insights into training SOTA video generation world models, covering Grok Imagine, Cosmos, and the parallels between video and coding agents.
xAI has released grok-build-0.1 in public beta via API, a model optimized for agentic coding tasks with high speed and competitive pricing.
Grok-Imagine-Video-1.5-Preview (720p) has secured first place in the Image-to-Video Arena, improving by 52 points over its predecessor and surpassing top video models like Seedance-2.0 and HappyHorse.
This paper investigates whether high-quality Natural Language Explanations (NLEs) generated by LLMs from XAI outputs actually improve task performance, finding they do not aid accuracy but inflate confidence, revealing a quality-usefulness gap.
A tweet highlights Grok Imagine's image and video generation capabilities, praising the high realism that blurs the line between AI and reality.
Elon Musk highlights the ability to paste screenshots directly into Grok Build, allowing users to visually explain ideas for inspiration, debugging, or recreation, streamlining the workflow.
A beginner video guide on how to install and use Grok Build, designed for non-technical SuperGrok and X Premium+ users.
Elon Musk announces that Grok Build, currently in beta, is already useful for production tasks, featuring fast web/X search, image creation with Imagine, and subagent/worktree integrations.