Claude Fable 5.2 is built for the hard, long work

Some work cannot be done in one quick prompt. Building a detailed financial model. Reviewing a large piece of software before launch. Checking a research claim against dozens of sources.

It needs patience, care, and someone who admits when they are not sure. This week, Anthropic released a model built for exactly that kind of work.

Claude Fable 5.2 is Anthropic’s most capable model for technical work, with a focus on coding, finance and science.

It finishes more real work

On a test built from real freelance projects, it completed 41.3% of them, up from 17.9% for the previous Fable model. On real coding tasks, it scored 69.9%, up from 56.3%.

It says when it doesn’t know

Cognition, a company that tested it early, says it “says it doesn’t know things instead of hallucinating,” which means making things up.

It costs less to build with

For companies building their own tools on it, Anthropic has cut the price by 20% compared with the previous Fable model.

For a business, this helps with the long jobs that take a person days. A finance team could use it to build a forecast. A tech team could use it to review code before a release. It works best with clear instructions, and with a person checking the result.

These are Anthropic’s own test results, so try it on one real piece of work before you rely on it.

Who gets it: Claude Pro, Max, Team and Enterprise users, from 9 November, with extra usage to try it. Developers can use it through Anthropic’s API and major cloud platforms. It is not on the free plan.

1A Pune startup is helping health workers find TB faster

Froncort.AI has received $2.3 million from Google.org. Its tool, AI-NETRA, uses health records, satellite maps and local data like rainfall and flooding to show health workers where to look first.

In a study with Maharashtra’s State TB Office, teams found about the same number of TB cases while screening around two-thirds fewer people. The cost per confirmed case fell from about $29,000 to $10,000, and 92% of health workers said their physical workload dropped.

Froncort was one of 15 organisations chosen worldwide for Google.org’s grant programme.

Who should care: anyone interested in how AI can help public health in India.

2Indian employees say AI helps. Now companies need to catch up.

A new Cognizant study surveyed 700 employees and 75 business leaders across 10 industries in India. 97% of employees say AI makes them more productive, with an average gain of 15%.

If companies turn those personal gains into business results, AI could add $28 billion to $84 billion to India’s GDP every year. The step that matters now is moving AI from individual habits into how the whole organisation works.

Who should care: business leaders planning how to roll out AI across their teams.

3ChatGPT can now make GIFs and remember your photos

ChatGPT has a new image model that is faster and better at keeping faces and product details the same when you edit. It can also turn an image into a looping GIF.

You can now save a photo, like your product or your logo, give it a name, and use it in any later chat by typing @ and that name. No more uploading the same picture again.

Who gets it: ChatGPT users worldwide, on all plans. GIFs are still rolling out, so they may take a few days to reach your account.

Until next Wednesday,

P.S. Know someone who saves posts but never goes back to them? Forward them this note.

Turn your saved X posts into your next post idea. It takes about 15 minutes. Saving posts on X is easy, but finding the useful ideas again is harder. Grok Bot can read your bookmarks, spot the themes and help you draft a post.

Final call: Cohort 3 starts this Saturday

Cohort 3 of our AI Workflow Accelerator for Professionals begins this Saturday, 14 November. Over four weeks of live sessions, you build practical AI workflows around the work you actually do, and leave with one you can use right away.

If you have been meaning to join, this is the week to do it.

5,000+ professionals

Founders, business leaders and practitioners working through the AI shift together, one conversation at a time.