OpenAI Made AI Free. DeepSeek Just Ended the Illusion.
OpenAI gave GPT-5.6 Luna to every ChatGPT user with no limits. DeepSeek raised its API prices by up to 12x effective tomorrow. Both moves reveal the same economics.
Original writing by Maxim — essays, analysis, field notes, and long-form thinking on AI, alignment, and building in the open.
145 results in this archiveOpenAI gave GPT-5.6 Luna to every ChatGPT user with no limits. DeepSeek raised its API prices by up to 12x effective tomorrow. Both moves reveal the same economics.
Three Central Asian animators finished a 15th-century nomadic epic in two weeks on Higgsfield's AI pipeline. It's the kind of film the industry's greenlight math reliably kills.
Z.ai's voluntary weight delay on GLM-5.3 is real — and the gap between fourteen days and Anthropic's indefinite restriction says everything about how differently these labs are reading the same capability.
Qwen3.8 ships the recall-conservative end of the validated hybrid ratio range, at a scale no study has tested. Its mandatory thinking mode sits beside a documented failure mode nobody has publicly checked.
Researchers turned one global encryption key into a cross-user chain-of-thought extractor, recovering 182 credentials from 315,000 publicly logged reasoning blocks.
Claude improved the known lower bound for zeros of the Riemann zeta function on the critical line from 41.6% to 67.2%. Conrey and Goldston read the proof. Here is what the result actually says.
I gave the plants in a small world a genome, and for most of a day they evolved in the wrong direction. Three times I designed the selection pressure, and three times the pressure I specified and the pressure I built were different objects. Only the measurement showed the gap.
Four webshells and six administrator accounts I did not create, from a WordPress core flaw I had published an article about five days earlier. The patch existed and was pushed to the whole internet. It never reached me, because four months before I had silenced the scheduler that installs patches while trying to make the site safer.
Oracle controls two major Java projects with opposite AI contribution policies. The difference isn't quality — it's liability.