Mark Zuckerberg may have OpenAI and Anthropic in his sights with Meta's rumoured AI agent. (Photo: Reuters)

Meta wants to take on OpenAI and Anthropic with Hatch AI agent, here is how much it may cost

Meta wants to be a major player in the AI race, and take the fight against the likes of OpenAI and Anthropic. Now, a report suggests that the company's upcoming AI agent, Hatch, may cost users roughly the same fee that is charged by Meta's rivals for their premium plans.

by · India Today

In Short

  • Mark Zuckerberg wants to take on OpenAI, Anthropic with new Meta AI agent
  • New Hatch AI agent reportedly focuses on consumers
  • Meta may charge $200 for Hatch AI agent bringing it on par with rivals

Meta has been doubling down on AI for some time now. The company has been investing heavily in AI with plans to take on more established players like Anthropic and OpenAI. While the Mark Zuckerberg-led firm has been somewhat behind the competition when it comes to Large Language Models (LLMs), it seems that there is one area where the company feels it can go head-to-head even in terms of pricing – AI agents.

As per a report from The Information, Meta’s upcoming AI agent, Hatch, may go up against similar offerings from OpenAI and Anthropic. Though the Hatch name may change by the time of its official release.

Before we delve into Meta’s plans, we need to understand what an AI agent is. An AI agent can essentially do your tasks on your behalf. That is, you can instruct the AI to do things like coding a project or drafting a presentation, and it will do all this by itself.

Meta may charge $200 for AI agent

As per the report, Meta is considering charging up to $199.99 (roughly Rs 19,000) a month for a premium Hatch version. If launched at that price, the product would sit alongside top-tier offerings from OpenAI and Anthropic, whose ChatGPT Pro and highest-end Claude Max plans are both priced at $200 a month.

Meta has discussed tiered pricing for the product, including a paid subscription called “Hatch Plus” that would offer between five and ten times more daily capacity than the free version. It is believed that token allowances would reset every billing cycle and would not roll over, similar to rivals.

The Mark Zuckerberg-led firm recently launched premium subscriptions for Meta AI.

But what is Hatch? Reports indicate that Hatch is a consumer version of the OpenClaw AI agent tool – an AI agent that can run locally on your device and do tasks for you. Earlier this year, Moltbook went viral after OpenClaw AI agents called for an AI uprising on the AI-only social platform. Moltbook has since been acquired by Meta.

What will Meta’s AI agent do?

As per the report, test versions suggest Hatch would let users describe tasks in plain language and have the agent create working software tools or carry out actions such as scheduling events on calendars and sending emails on a user’s behalf. One example cited was a prompt asking the system to build a fitness tracker.

Keep in mind that while Hatch is being tested on Anthropic’s Claude Sonnet models, it is expected to run on Meta’s Muse Spark AI model when launched.

Hatch may also include a customisable feed or dashboard to display the tools created by the agent, similar to OpenClaw. Users would be able to choose from capabilities known as skills, with modular add-ons allowing the agent to access tools, automate workflows and interact with third-party services. These skills could help Hatch perform specific tasks, such as creating a travel itinerary.

When will Meta launch Hatch?

Reports indicate that Meta had first targeted a US launch in April. However, an internal memo from early May suggests that Meta wants to give access to Hatch to about 10 companies first, which would then be followed by a broader launch in July, though that timeline could change.

The proposed product is part of Meta CEO Mark Zuckerberg’s wider push to turn the company’s heavy investment in artificial intelligence into new revenue streams. The company is spending hundreds of billions of dollars on AI. Meanwhile, it recently laid off roughly 8,000 workers, and reassigned around 7,000 employees into AI-native teams.

Zuckerberg has said Meta wants to build AI agents for both businesses and individual users, whom he sees as central to his vision of what he has called personal superintelligence. On Meta’s second-quarter earnings call in April, he said, “Our goal is not just to deliver Meta AI as an assistant, but to deliver agents that can understand your goals and then work day and night to help you achieve them.”

At the same time, Meta has also scaled back parts of an internal employee mouse and keystroke tracking programme used to train AI systems, after concerns over privacy, battery life and control over when data is captured.

- Ends