Grok Bot adds text- and image-to-video generation

• Grok Bot 0.75.2 adds video generation from descriptions or images, as the launch announcement details.
• The bot can assemble a complete edit: venturetwins reports that it chose a storyline, found and stitched clips, and added titles and music.
• Wider access is still pending: ericzakariasson said he'd post when the rollout goes further.

A 71-second self-portrait is the standout demo, with the bot making the editorial decisions. The release notes also tuck in another creative feature: tasks handed off by the bot can generate images and videos for the chat.

The official changelog dates version 0.75.2 to October 22. For image animation, ericzakariasson's demonstrated prompt is just: turn this into a video.

Asked to make a video about itself, the bot handled five production jobs, according to venturetwins:

1. Set the storyline.
2. Found clips.
3. Cut and stitched them together.
4. Added titles.
5. Added music.

That account leaves the clip sources unspecified, so it establishes an editing workflow without identifying where every shot came from.

moritzkremb's 24-minute walkthrough shows how he organizes ongoing creative and business work:

• A chief-of-staff bot sits above specialist bots, including content bots.
• A partnerships bot reads incoming brand requests and tracks sponsorship deals in Notion, rather than relying solely on built-in memory.
• A project bot starts local or cloud tasks in Cursor, with an explicit model choice demonstrated in his coding walkthrough.

Two companion features ship in the same release:

• Pages: reports, memos, and dashboards that users edit with the bot or have it keep current.
• Desktop chat cards: redesigned approvals, questions, files, links, and tables.

Pages can update proactively without overwriting work in progress, ericzakariasson said in a reply.

An illustrative inventory Page configured to refresh every morning.

ericzakariasson said he would post again when the release rolled out further, without giving a date.

Neither the video model nor per-video pricing is specified in the 0.75.2 release notes.