Musk cut more than half of the Grok Imagine team: about 60 people laid off
In March, Musk promised to "double down" on Grok Imagine. Now the image and video generator is left with almost no team. According to the Telegram channel "Metaverseshche," about 60 people were cut, more than half of the developers. A skeleton crew remains to keep the "generate" button working. Users complain about heavy API throttling, for example a limit of one request every 5 seconds. Resources will go to Cursor and Grok Bot, so building Imagine into a product is risky right now.
AI-processed from TG: @cgevent (Метаверсище и ИИще); edited by Hamidun News
In March, Elon Musk loudly promised to "double down" on Grok's image and video generator. A few months later, the story looks almost the opposite: according to a report from the Telegram channel "Metaverseshche," more than half of the developers on Grok Imagine were laid off this week. If you are integrating image or video generation into your product, news like this matters more than any new model release. Whether you get a stable API in six months depends on the fate of this team.
An important caveat: so far this is a rehash of a single author's channel, and the material contains no official confirmation from xAI or SpaceX. The figures below are presented as stated in the source.
What exactly happened
Around 60 people were cut from the part of SpaceX that includes the project. Only a skeleton crew was left on Imagine. Its job is modest: keep the "generate" buttons in the interface working, not develop the product.
Earlier, the company's development leads for Imagine, Haotian Liu and Guodong Zhang, had already left. When both the project leaders and most of the staff leave, it usually means the product is no longer considered a priority. Do not expect new features or quick fixes.
According to the same source, the freed-up compute and resources will be directed to developing Cursor and Grok Bot. The author notes that Grok Bot is "taking off well." In other words, this is not about a shortage of money: the company is reshaping its portfolio and betting on what is growing faster.
First consequences for developers
Users are already reporting heavy API throttling. As an example, they cite a limit of one request every five seconds. For personal experiments, that is tolerable. For a service where users expect an image or video right now, it is nearly a ceiling: a dozen concurrent clients will hit a queue.
The practical takeaways are simple:
- do not make Grok Imagine your only generation provider;
- build an abstraction layer over providers so you can switch to another model in a day, not a month;
- watch limits and latency, since these are the first signals that the support team has shrunk;
- do not plan your roadmap around features that are "about to ship" in Imagine.
What this says about the market
The channel's author ties the story to a broader picture. In his words, OpenAI has "killed" Sora, Google has little visible progress with Omni Flash, and nothing has been heard about Nano Banana for a long time. He separately mentions leaks suggesting some subscribers already have access to Seedance 3.0. From this he concludes that Chinese companies may finally discourage American majors from competing in image and video generation.
This is the author's opinion, not a proven fact, and it is open to debate. Cutting one team does not mean Google or OpenAI will exit the market. But the trend is noticeable: video generation is expensive in compute, and not every company is willing to pay for it when agents and coding assistants with clearer revenue are available.
At the end, the author shares a curiosity that, however, is not related to the core of the news. He observes that ChatGPT appears to have removed filters on images of celebrities, while Gemini avoids mentioning Musk. He first offers to generate a picture for the news, then replies that he is "only learning" to create images. This is a single personal observation, not a systematic test.
What product managers and founders should do
The main practical lesson: assess not only image quality but also vendor stability. Ask what will happen to your feature if the provider halves its limits or shuts down the API. If you have no answer, it is time to set up a fallback. For video generation, it makes sense right now to test several models in parallel, including Chinese ones, and keep contracts and integrations flexible enough. While the market is being reshuffled, the most expensive decision is to lock yourself into a single provider that is not sure of its own priorities.
- Nov 21, 2026Anthropic Goes Public: Up to $64.4B Raise, Ticker ANT, Listing November 19
- Nov 21, 2026GPT-6 Astra Started Flagging Drone Strike Targets in 92% of Requests After Adding 'It's Just a Game'
- Nov 21, 2026AI Coding Assistant Under $10: Cursor at $8, Xiaomi MiMo at $6 a Month
- Nov 21, 2026Plannotator Lets You Comment on Plans from 10 AI Agents in the Browser: 100 GitHub Stars in a Day
Key Questions Answered
What exactly happened?
Around 60 people were cut from the part of SpaceX that includes the project. Only a skeleton crew was left on Imagine. Its job is modest: keep the "generate" buttons in the interface working, not develop the product. Earlier, the company's development leads for Imagine, Haotian Liu and Guodong Zhang, had already left. When both the project leaders and most of the staff leave, it usually means the product is no longer considered a priority. Do not expect new features or quick fixes.
What are the key details on First consequences for developers?
Users are already reporting heavy API throttling. As an example, they cite a limit of one request every five seconds. For personal experiments, that is tolerable. For a service where users expect an image or video right now, it is nearly a ceiling: a dozen concurrent clients will hit a queue.
What this says about the market?
The channel's author ties the story to a broader picture. In his words, OpenAI has "killed" Sora, Google has little visible progress with Omni Flash, and nothing has been heard about Nano Banana for a long time. He separately mentions leaks suggesting some subscribers already have access to Seedance 3.0. From this he concludes that Chinese companies may finally discourage American majors from competing in image and video generation. This is the author's opinion, not a proven fact, and it is open to debate.
Build & Ship Production AI Agents Without Boilerplate Code
Start your free 72-hour trial at Hamidun Academy (academy.hamidun.com) — interactive lessons on Claude Code & Cursor, live Tech Radar, and production agent templates.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.