This week the AI race got more interesting for the rest of us. Anthropic put out a better Claude and started peeling away paying users from ChatGPT, while the White House asked OpenAI to slow down. Here is what actually matters for your work.
ANTHROPIC
Claude Opus 4.8 is here, and it is built for real work
Anthropic released Claude Opus 4.8, an upgrade to its top tier of models. The headline gains are in coding, agentic tasks (where the AI carries out multi-step jobs on its own), and professional work like writing and analysis.
The part worth noting is consistency. Anthropic says this model holds up better on long-running tasks, the kind that take many steps and used to drift off course halfway through.
For builders, that means you can hand Claude bigger jobs and trust it to finish them without constant babysitting.
you can hand Claude bigger jobs and trust it to finish them without constant babysitting
Why it matters A more reliable model means you can automate longer, messier tasks instead of just quick one-offs.
ANTHROPIC
People who pay for AI are picking Claude
ChatGPT still has the biggest crowd by far. But new data shows that among people willing to pay for AI, more and more are choosing Anthropic's Claude.
That is a meaningful signal. Free users try everything; paying users vote with their wallets for the tool they trust to do serious work.
If you have only ever used ChatGPT, this is a nudge to test Claude on your hardest tasks and see which one fits how you actually work.
Free users try everything; paying users vote with their wallets for the tool they trust to do serious work
Why it matters The tool serious users pay for is usually the one worth your time too.
OPENAI
The White House asked OpenAI to slow down
OpenAI reportedly plans to share its newest model, GPT-5.6, with a small group of partners instead of releasing it to the public. The reason, according to the report: the Trump administration asked it to, citing safety concerns.
This is unusual. A government stepping in on a model launch signals that the most capable AI is now treated as something closer to critical infrastructure.
the most capable AI is now treated as something closer to critical infrastructure
Why it matters When the most powerful models stay behind closed doors, the tools you can actually access may lag the headlines.
RESEARCH
GPT-5 helped crack a 3-year science puzzle
Immunologist Derya Unutmaz had been stuck on a mystery about T cell behavior for three years. GPT-5 Pro helped him find an answer, offering insights that could feed into cancer and autoimmune research.
This is a good example of AI as a thinking partner, not just a writing tool. It did not replace the scientist. It helped him see what he had been missing.
It did not replace the scientist. It helped him see what he had been missing.
Why it matters AI is becoming a real collaborator on hard problems, which is a higher bar than drafting emails.
WORK
Agents are starting to do longer jobs
A new OpenAI research paper describes how AI agents are changing work by handling longer, more complex tasks across many different roles.
The shift is from quick answers toward AI that can carry out a whole process. Think of it less like a search box and more like a junior teammate you can delegate to.
The takeaway for small teams: the next productivity jump comes from handing off entire workflows, not single prompts.
Think of it less like a search box and more like a junior teammate you can delegate to
Why it matters If you can describe a repeatable process, you are getting closer to handing it to an AI agent.
Hit reply and tell me which one won your test, Claude or ChatGPT. I read every response and the best ones shape next week's issue.
Free · every Friday
Get the weekly AI brief
Plain-English AI, in your inbox each week. One email, no spam.
No spam, ever. Unsubscribe with one click. Phone is optional and only used if you ask us to call.
Prefer the website? Subscribe at openhour.io →
