Short reports with sources and follow-ups. Checking a source does not confirm a rumor; each story’s status is shown separately.
Browse all storiesShowing 13 stories
A new model label has surfaced before a confirmed launch. The screenshots are the clue; claims about a rollout and better performance go further than the evidence.
A merged Qwen Code change prepares for the next model family. The code is real; the model’s capacity and launch date remain unconfirmed.
The Information reports a physical prototype and work on a second version. Reuters relays the account but says it could not immediately verify it.
A report names a December window for consumer reservations. Musk’s one-word reply pushes back, but leaves the details unresolved.
A new claim points to internal 32B, 122B and 256B models. The author says those numbers describe training work, not a downloadable lineup.
Three words in an official animation have acquired a next-model theory—and a much stronger claim that GPT-6.2 made the entire film from code. The frame is real. That production story is unverified.
A bicycle-riding pelican, followed by an animated SVG, accompanies a claim that “Fable 5.5 Max” is coming to Claude Code. The author has shown the result, but not the model behind it.
TestingCatalog shared an alleged developer-only screen for background tasks. Google has confirmed Jules’ shutdown, but not a release date for this interface.
A reported server-delivered instruction targets recreations of specific existing games. It does not establish a ban on making games with Claude.
An alleged tracker entry has given the Bel story a new name. The link to GPT-6.2 is still a claim, not an OpenAI announcement.
Musk is promising a more useful Grok. The missing detail is when people will get it.
The claim circulating on X has a real source: Google’s earnings-call transcript. What it establishes is narrower than an autonomous AI breakthrough.
An image-editing parameter names the model. A public launch remains unconfirmed.
