Anthropic Readies Model That Reads Users' Intentions Better

In the last month, OpenAI, Anthropic and Google have released more than a dozen new models. If you thought that pace was crazy, don’t expect things to slow down anytime soon.

Anthropic is preparing to release a new model that is better in its reasoning and intuition than the Anthropic models that are currently available commercially, according to a person who has used 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, such as explicitly stating those goals, the person said.

For instance, a person might ask the model to fix a single bug, and the model might go off and instead identify the underlying class of problems that caused the bug and ask whether it should fix the whole class of problems, said the person who has used it.

The Anthropic model is also better at thinking outside of the box rather than regurgitating common sense advice or solutions, the person said. For instance, if a user asks for advice on how to invest in stocks, rather than providing generic advice like looking at price-to-earnings ratios, the model might come up with its own creative algorithms to predict future stock performance, the person said.

It’s not clear which model family the Anthropic will fall into, such as Fable (the biggest and most capable model) or Opus (the second most capable one).

Meanwhile, Google is also working on another model, codenamed Gemini 4 Carbon, a person with knowledge of the model said. (Business Insider earlier reported the name of the new model.) The company released its last model, Gemini 4 Argon, last week, and Carbon appears to be a step up from that one, the person said. Though Argon didn’t appear to leapfrog Anthropic or OpenAI’s most advanced models, some developers tell me that Gemini models continue to be much better at tasks that require visual understanding, like analyzing videos or testing that a website works properly. (Perhaps it helps that Google has developed models with data from YouTube videos!)

Last but not least, OpenAI is also working on a new model that will likely be released in the upcoming weeks, said a person with knowledge of it. This model will likely be the newest iteration in OpenAI’s “Terra” model family, its mid-tier option that sits between its “Sol” and “Luna” models, they said. This model will be an improvement in capabilities, but will be cheaper than GPT-6.1 Sol, the model OpenAI released late last month that’s almost as good as its flagship Astra model but at a lower price, the person said.

This next release underscores OpenAI’s recent focus on undercutting Anthropic, the market leader, on cost, as well as its push to release models at a variety of price points and capability levels to give developers more options.

However, OpenAI is still in an all-hand-on-deck mode to prepare the next major, unreleased model (which has some looping transformer components) for release without major safety or security risks. That could take some time, we reported last week.

The plethora of model options these days is exciting to application developers. But, personally speaking, it’s making me nostalgic for the olden days of 2025, when reporting about one or two model releases per quarter already felt overwhelming. And, it’s also highlighting how a growing number of businesses and developers aren’t automatically reaching for the most advanced models.

Here’s what else is going on…

Overheard

Microsoft has pulled out of a planned U.K. data center Nscale was building for it, after the cloud provider faced delays accessing power to bring the site online. The cancellation is likely to weigh on Nscale’s planned public offering this month as investor concerns swirl over whether AI cloud providers can bring in revenue fast enough to pay their debts. Firmus Grid, an Australian competitor to Nscale, abruptly pulled its planned IPO earlier this month, and shares in AI cloud provider CoreWeave fell more than 23% last week.

China is considering new export controls on AI technologies that could potentially restrict the overseas transfers of AI models and chip designs deemed strategically important. The proposed curbs cover 23 AI-related technologies such as large language models, AI agents, chip design and computing clusters.

Over the weekend, Anthropic announced the creation of a new fund that will get at least 5% of its shares and support public-benefit causes such as addressing the employment impact of AI. Its Claude models also solved a decades-old puzzle of the original algorithm that had been used to construct one of the most popular cryptographic standards.

Deals and Debuts

See The Information’s Generative AI Database for an exclusive list of private companies and their investors.

Anthropic is offering Blackstone a yield above 10% to anchor a $9 billion junior slice of a $60 billion debt package the AI company is assembling to buy Google-developed TPU chips, according to the Financial Times.

Proofpoint, a portfolio company of Thoma Bravo, agreed to acquire AI agent security company Varonis Systems for approximately $6.3 billion in an all-cash deal.

GTCR and Elliott Investment Management agreed to acquire CCC Intelligent Solutions, a Chicago-based SaaS and AI platform for the insurance and auto-repair industry, in an all-cash deal with an enterprise value of approximately $6.1 billion.

Intuit agreed to acquire Sequence, a London- and New York-based Series A company whose AI-native platform automates billing, pricing and revenue recognition for mid-market businesses, in a deal reported at roughly $200 million.

Abby Care, which uses AI to train paid family caregivers, raised $82 million in a Series C funding round led by Transformation Capital.

Stand, which uses AI and physics simulation to price wildfire risk, raised $70 million in a Series C funding round led by Saga Ventures.

North.cloud, which uses AI to track cloud and AI spending, raised $60 million in a Series B funding round led by PeakSpan Capital.

Holyvolt, which uses AI to discover new battery materials, raised $45 million in a funding round backed by Zone II and CC.VC.

Sirona Medical, which makes cloud-based AI radiology software, raised $40 million in a Series D funding round led by Avidity Partners.

Sunbound, which uses AI to automate senior-care billing, completed its acquisition of Richter Healthcare Consultants after Sunbound raised $24 million in a Series A funding round led by TTV Capital.

Opsin, which secures enterprise AI agents, raised $21 million in a funding round led by SYN Ventures and Race Capital.

Origin, which builds AI robots for construction, raised $20 million in a funding round co-led by Accel and Zacua Ventures.

CarbonAI, which uses AI agents for commercial real-estate lending, raised $12 million in a funding round led by Cowboy Ventures.

Goodweek, which helps companies manage AI tool access, raised $12 million (€10.69 million) in a seed funding round led by Adara Ventures.

Chariot Claims, which uses AI to file consumer legal claims, raised $10 million in a Series A funding round led by Silverton Partners.

Elorian, a Palo Alto AI startup founded in 2025, released its first model Glimpse on Friday. Elorian says the model reasons natively in images, not just language or code.

Anthropic began embedding AI evaluators within the company to assess and publish reports on safety practices and risks, the company announced in a blog post last Thursday.

DeepSeek’s latest update to its affordable lightweight model, V4.1-Flash, is making a big splash in the industry because of significant improvements from the previous version that bring it closer to frontier models.

Nvidia has asked the suppliers of its next-generation custom high-bandwidth memory chip to use Taiwan Semiconductor Manufacturing Co’s 3-nanometer manufacturing process, the Seoul Economic Daily reported.

Meta launched Muse Spark Open, an open-weight model that’s able to run on Macs.

Thank you for reading the AI Agenda Newsletter! I’d love your feedback, ideas and tips: [email protected].

If you think someone else might enjoy this newsletter, please pass it forward or they can sign up here.