Image: Anthropic

Anthropic is getting ready to release a new Claude model that is better at working out what you actually want, even when you don't spell it out. That's according to The Information, whose reporter Stephanie Palazzolo says the model could arrive this month, citing a person familiar with it.

Expect this month to be a busy one for new model releases. We have the scoop on a new Anthropic model, Google’s Gemini 4 Carbon, and a potential new Terra model from OpenAI.

— Stephanie Palazzolo (@steph_palazzolo) October 19, 2026

The report is paywalled, but TestingCatalog shared key lines from it. "The unreleased model is especially good at discerning a person's underlying intentions and goals even if they don't give the model much information," the report says. It adds that the model is "better at thinking outside of the box rather than regurgitating common sense advice or solutions."

In plain terms, "reading intentions" means filling in the gaps. Most people type short, vague requests. Today's chatbots often answer the literal question, or reply with safe, generic tips, and you end up writing three follow-ups to get what you meant. A model that infers the goal behind the question could get there on the first try.

The Information gives an example, according to a screenshot TestingCatalog posted. Ask how to invest in stocks, and instead of the usual advice about price-to-earnings ratios, the model "might come up with its own creative algorithms to predict future stock performance." Finwire, which summarized the report, says the same applies to coding. When fixing one bug, the model can reportedly spot the wider set of problems behind it and suggest a broader fix.

That last part may matter most to everyday users of tools like Claude Code. Fixing the symptom you pointed at, while the real cause stays hidden, is one of the most common complaints about AI coding help.

ANTHROPIC 🔥: According to The Information, Anthropic will release a new model this month, along with Gemini 4 Carbon and a new Terra model from OpenAI.

— 🚨 AI News | TestingCatalog (@testingcatalog) October 19, 2026

What the model will be called is still unknown. "It's not clear which model family Anthropic will fall into, such as Fable or Opus," the report says. Anthropic hasn't commented publicly, and there are no benchmarks or hands-on tests yet, so the claims come from a single unnamed source for now.

Still, the timing lines up with other signs. Claude Fable 5.5 is the only member of Anthropic's 5.5 family that hasn't launched. We reported it showing up in Microsoft's Foundry catalog and reaching Anthropic's partners earlier this month. Testers have also reported an unnamed new model answering in Claude Code.

There's a business reason to move quickly, too. A month ago, Reuters reported that Anthropic was weighing a new model release to answer OpenAI's GPT-6 Astra ahead of its expected IPO. At the time, Astra had about 13% of the business AI spending tracked by Ramp, against 8% for Claude Fable. A model that feels noticeably smarter from vague prompts is exactly the kind of jump that wins back paying customers.

Anthropic won't be alone. The same report says Google is preparing Gemini 4 Carbon, and Finwire says OpenAI's next Terra model could land within weeks, at a lower price than GPT-6.1 Sol.

One caution: a model that guesses your intentions can also guess wrong. TestingCatalog called the "underlying intentions" line "a bit scary," and how well Anthropic balances helpful guessing with asking before acting is something to watch once people can try it.