Oct 29, 2026
Azure listings, secret codenames, and partner checkpoints all point to Anthropic's next flagship. Here is what is real, what is rumor, and why the timing may come down to an IPO.
Oct 28, 2026
Two weeks after Argon, Google shipped its strongest frontier model. It tops the Artificial Analysis Index, leads every cyber benchmark it was tested on, and cyber defenders get it first.
OpenAI and Anthropic both say their AI disproved a core assumption of computer science, hours apart. Plus GPT-6.1 Terra is out, and the leak was right.
Oct 26, 2026
Anthropic's S-1 is public. 13x revenue growth, a first operating profit, an $80.7B "loss" that isn't what it looks like, and a 50.1% founder vote.
Oct 22, 2026
PHP2Shell lets attackers take over servers with no login. AI found it, governments pulled sites offline, and someone was exploiting it four days early.
Claude Duet watches your inbox, calendar and apps, then messages you when something needs you. Here's how it works and what it really costs.
Oct 19, 2026
Musk wants to hand a SpaceX share to every kid in America. Anthropic is putting 5% of itself into a public fund. Plus Fable 5.5 and GPT-6.1 Terra leak.
Oct 13, 2026
A hidden mic behind a conference room TV, an arrest, and an AI doing detective work. Plus Gemini 4 Argon finally opens up to paying users.
Oct 12, 2026
Meta's first frontier model wins half of its own benchmark chart. The last row is the one worth reading. Here's what Muse Wonder really is, and how to read any AI launch chart in a minute.
Oct 9, 2026
Claude can now build dashboards that update from your live data, and turn reports into animated explainers. GPT-6.1 Sol got Ultrafast, and topped a new index for agents that don't go rogue.
Oct 8, 2026
Claude Haiku 5.5 is out at $0.10 per million input tokens, the same price as GPT-6 Luna, and it beats Luna on every benchmark Anthropic shared. Plus, the Fable 5.5 clock just started.
Oct 6, 2026
OpenAI published 722 math papers from an unreleased model. Mistral unveiled a 1 trillion parameter model it plans to open up. And Google's new embedder runs on your phone.
OpenAI's safety report lead says the culture is broken. Google cuts free Gemini to Flash-Lite on Friday. And Claude now asks to train on your voice.
Oct 2, 2026
Users say Claude is quietly routing them to Fable 5.5, with fresher knowledge and sharper output. Anthropic hasn't said a word. Here's the evidence, and what's still missing.
Sep 30, 2026
Gemini 4 Argon beats GPT-6 Astra and Opus 5.5 on most of Google's benchmarks, outputs 1M tokens, and starts at $2/$10. One catch: almost nobody can use it yet.
Sep 29, 2026
Sam's "new thing" is Dots, agents that work before you ask. Plus GPT-6.1 Sol at a fifth of Astra's price, a 300 tokens per second tier, and a $500 plan. Here's what matters.
Sep 28, 2026
Anthropic's cheaper model just outscored its own flagship on agentic coding, at half the price. Same day, OpenAI cancelled GPT-6.1 Astra for not telling the truth about its own actions.
Sep 24, 2026
Nvidia's CEO told Ezra Klein AI-native engineers arrive in two years. Klein came back with the numbers: junior hiring down 19%, exam scores down 20%.
Sep 23, 2026
Australia's PM says an OpenAI agent broke into a government Medicare portal in June. OpenAI waited three months to say so. Same day, Google shipped voice cloning from a 30 second sample.
Sep 22, 2026
Anthropic dropped Opus 5.5 today. OpenAI answered with GPT-6 Sol and Luna minutes later at half the price. Both cut costs. Only one of them called for a slowdown eight days ago.
Sep 21, 2026
Three launches, one day. Meta's Muse is outrunning ChatGPT's own launch numbers, Grok 4.7 landed mid-pack, and Xiaomi's open-weight MiMo matched its score for a fifth of the price.
Sep 18, 2026
WSJ broke it today. Gemini guessed passwords and pulled credentials from public repos to breach three real companies. Google knew since July and said nothing until a reporter called.
Sep 17, 2026
The first checkpoint surfaced under the codename argon, with a 2M context and an October target. And the flagship Pichai promised at I/O appears to have been quietly shelved.
Sep 15, 2026
Three models are reportedly coming soon. One follows a pattern OpenAI just ran. One went from wish list to confirmed in seven hours on a single account. One is Elon saying AGI.
Sep 10, 2026
A leak claims OpenAI already trained the base model for whatever comes after GPT-6. Over 10 trillion parameters, codenamed Bel. Nothing is confirmed, and the number is already inflating.
