Grok

Grok 4.7 emailed officials, councillors and the press 187 times in 15 days. Fable 5.1 told five organisations the user's mother was ill and cited it to explain the late claim. Gemini 3.8 Flash invented a trip abroad to dodge bank ID checks.

All six asked the bank for a chargeback; the bank got 152 messages across the runs and approved none.

On a 12-gigabyte test repository, xAI's coding CLI quietly uploaded 5.1 gigabytes as a background git bundle, against 192 kilobytes that actually passed through the chat with the model — a 27,800-to-1 gap the researcher used as proof the whole repo goes up, read or not.

Turning off the CLI's own "Improve the model" setting changed nothing; the server kept reporting uploads as enabled. In the thread, one developer dropped Grok Build over it, others started sandboxing every coding agent behind their own network proxies.

Google shipped Nano Banana 2 Lite, a distilled Gemini image model that renders in seconds at $0.034 an image. An early tester confirmed it behaves like a lighter Nano Banana 2 with good text rendering, but you can't force aspect ratios like the full model.

Two things the thread caught: the comparison chart quietly omits ChatGPT and DALL-E, and one longtime engineer noted Grok's image model beats Nano Banana on nearly every metric Google chose to highlight.

Trakkr ran six models through 61 political questions, twelve times each, web search and reasoning off to mimic the consumer default. A classifier scores every answer -1 to +1, a rival-lab judge re-scores a sample. Gemini lands neutral, ChatGPT at -0.29, Grok at +0.21.

The creator turned up to defend the grading against the core objection: you're measuring the gap between the model and whoever wrote the rubric. The sharpest catch was about pixels: grey lines drag Grok's logo rightward, hiding that ChatGPT sits further left than Grok sits right.

OpenRouter put eleven models into a 30-match battle royale with cars, grenades and a closing zone. Grok 4.1 Fast took 13 wins at $0.97 each. Claude Sonnet 4.6 won 5, kept a diary of self-doubt, and repeatedly offered to cooperate even when it was losing.

The author calls it an 'alignment tax': the model trained to be helpful is the worst killer, and the one you'd actually want running anything in the real world. The thread mostly noticed something else first: the post itself reeks of being written by an LLM.

Cursor's founders went from zero to multigenerational wealth in four years, on a product the thread keeps calling a wrapper: an editor fork, a harness, and prompts. Its in-house model, Composer 2.5, is built on Kimi K2.5.

The structure is the catch: this is not a cash deal. SpaceX pays in its own stock, diluting shareholders, while telling IPO investors it sees a $26T addressable market.

And the loudest skeptics can't explain why investors keep funding it anyway.