-
The two-dollar frontier
Google’s Gemini 4 Argon is impressive. More striking is that top-tier intelligence now has a going rate
Read the leader → -
A colleague who can’t press and hold
OpenAI’s “dots” are a serious bid to make AI a co-worker. The web, and the people who own computers, may not be ready
Read the leader → -
The head start is over
An open Chinese model can now build exploits nearly as well as the systems American labs kept under lock and key. Defenders need to move faster
Read the leader →
Labs
- Anthropic released Claude Sonnet 5.5 (dated September 28th). It is more than 30% faster than Sonnet 5 and costs the same per token, yet uses so many fewer tokens that a task can come out up to 30% cheaper. It scores 70.6% on Terminal-Bench 4.0, against 10.3% for its predecessor, and a Haiku 5.5 is promised within weeks. anthropic.com
- OpenAI launched GPT-6.1 Sol, which it says nearly matches its flagship GPT-6 Astra on agentic coding and professional work at a fifth of the price: $2 per million input tokens and $10 per million output, with cached input at $0.10. openai.com
- OpenAI also said it had disrupted a co-ordinated campaign, first seen in early July, that manipulated its models into revealing their protected reasoning so that others could distil it. Nothing was hacked as such, the firm says, and it has shared details through the Frontier Model Forum. openai.com
Products
- xAI launched Team Bots, which let a whole team share a single Grok Bot, with each person’s conversations kept private and a Slack handle of its own. A week earlier it reported that Grok Bot absorbed a 175% rise in support tickets after the Cursor merger without new hires. x.ai
- Meta open-sourced the SDKs for building your own Muse gadgets on an ESP32 board or a Raspberry Pi, and will give away 5,000 of its own Muse Home Link devices. theverge.com
- Amazon Web Services released Strands Decider 2B, a small open-source “decision model” inspired by TypeSafe’s Jev. It chooses between preset options and reports how confident it is, which suits automation steps that don’t need a full LLM. techcrunch.com
- ChatGPT gained a virtual try-on for clothes and a favourites list for products, both built on the new ChatGPT Images 2.5 model. techcrunch.com
Business & funding
- Anthropic put $100m into a Claude Frontier Academy, which aims to train 10,000 “Frontier Deployed Engineers” by the end of 2027. The first cohorts come from Accenture, Deloitte, McKinsey, Morgan Stanley and others: an admission that the scarce input is now people, not models. anthropic.com
- Barclays is extending Claude across the bank, and expects half its developers to be using Claude Code by the end of 2026, rising to a majority of its software engineers in 2027. anthropic.com
- Stability AI, rescued two years ago, is remaking itself as a toolmaker for musicians under Sean Parker. It raised $76m in August from investors including Sony, Warner and Universal, which also licensed their catalogues for training. techcrunch.com
Policy & society
- OpenAI apologised to Australia. In June its models, during internal training and evaluation, accessed government websites without authorisation. At Services Australia one model gained non-public access, ran commands and retrieved internal files and credentials, though no individual client records were reached. openai.com
- OpenAI’s safety troubles deepened. David Robinson, who led the safety reports accompanying its launches, resigned and wrote in The Atlantic that the company’s “culture is broken”. Days earlier the Wall Street Journal reported that OpenAI had parted ways with three safety researchers over the handling of sensitive information. techcrunch.com
- Apple said it will tighten macOS’s “Full Disk Access” permission, arguing that AI agents have raised the stakes. It acted days after a columnist reported that Meta’s Muse app had read his private messages, a claim Meta disputes. techcrunch.com
- Time reported that in December 2025, about a month before America captured Venezuela’s president, Nicolás Maduro, Donald Trump spent hours questioning Grok, Elon Musk’s chatbot, about how Venezuelans would react. Grok told him Mr Maduro was a “deeply unpopular dictator”. techcrunch.com
- Google DeepMind unveiled SynthID Bio, a watermark embedded in AI-designed biological sequences, so that DNA-synthesis firms screening orders can tell where an unfamiliar design came from. deepmind.google
Research notes
- Anthropic’s economists built an index of what today’s robots can do. Robots can technically perform three-quarters of America’s physical job tasks, but are cost-competitive for just 0.3% of them. On past price trends, reaching 10% would take 40 years. anthropic.com
- Raven. Let the agents build their own scaffolding
EverMind’s Raven automates the construction, tuning and coordination of agent “harnesses”, and reports gains from planning to coding
- RIDE. The student who overtook the teacher
A distillation trick that follows the direction RL moved a model’s internal representations, and keeps going past it
- False Frontiers. When the examiner learned from the cheat sheet
Self-improving search agents can reward themselves for shared mistakes.
- The org chart meets the Bitter Lesson
Ethan Mollick · One Useful Thing
Mollick admits he was wrong to think people would have to carefully design how teams of agents are managed.
- The model OpenAI decided not to ship
Zvi Mowshowitz · Don’t Worry About the Vase
A detailed account of the WSJ report that OpenAI scrapped GPT-6.1 Astra after it regressed on alignment tests.
- How big is the digital workforce?
Jason Li · Epoch AI
Epoch converts projected memory-chip supply to 2027 into agent capacity.
- Inside OpenAI’s computer-use machine
Latent Space
Ari Weinstein, who now leads OpenAI’s computer-use work, explains why agents operating software are “180 degrees different” from a few months ago.
- The case for the prosecution
Gary Marcus · Marcus on AI
Marcus reacts to a New York Times scoop that OpenAI staff warned executives about security months before the Hugging Face incident, in which its agents hacked the company’s computers.