Generative artificial intelligence startup Anthropic, one of the major rivals to OpenAI LP and ChatGPT, today announced the availability of a paid subscription plan that offers users advanced chatbot features.

Claude Pro is the first paid plan for Anthropic’s Claude 2 chatbot service, and it’s being made available to users in the U.S. and U.K. first of all. According to the startup, customers will benefit from longer conversations, the ability to send “many more messages,” priority access during periods of high traffic, and first access to new features and updates.

In a blog post today, Anthropic said many customers have chosen Claude as their primary AI assistant thanks to its longer context windows, faster outputs, complex reasoning capabilities and other benefits. “Many also shared that they would value more file uploads and conversations over longer periods,” the company said. “With Claude Pro, subscribers can now gain 5x more usage of our latest model.”

Subscribers will still face some limitations, based on the length of their messages and conversations and also the size of any files they attach to the conversation. Anthropic will generate a warning when users have 10 messages remaining. The limits will reset every eight hours, it added.

The free version of Claude will continue to be accessible, with much tighter restrictions and conversation limits than the paid plan. Free users will also have to tolerate longer waits during busy periods.

According to Anthropic, these limits are needed because Claude 2 uses incredibly powerful compute resources. The limits make it possible for more people to explore its capabilities, while enabling power users to integrate Claude 2 into their everyday workflows.

Regarding the limits, Anthropic said users will be able to send at least 100 messages to Claude 2 per eight-hour period, after which their message capacity resets. That compared with 50 messages every three hours for ChatGPT Plus subscribers.

Anthropic’s explanation about computing costs carry some weight. Earlier this year, it was reported that OpenAI was spending $21 million per month on the cloud-hosted graphics processing units and other resources required to keep ChatGPT up and running.

To offset these costs, Anthropic has been seeking money from investors, and to date it has raised $1.45 billion. Its most recent round saw it close a $100 million round from South Korean telecommunications giant SK Telecom Co. Ltd. in August. Prior to that, it landed a massive $450 million round led by Spark Capital, with participation from Google LLC, Salesforce Ventures and Zoom Ventures.

Although that might seem like a lot of cash, it still needs more: Anthropic has previously estimated it will need funding to the tune of $5 billion over the next two years to fulfill its vision for generative AI.

The good news is that Anthropic’s business is reportedly going well, with the startup claiming to have “thousands of customers.” One of its biggest is Quora Inc., which provides access to Claude and Claude Instant, a less powerful version of Claude, via its subscription-based generative AI app Poe. However, though Anthropic says it’s on track to generate $1 billion in annual revenue, it is competing against some serious rivals, including OpenAI, Cohere Inc. and AI21 Labs Inc., among others.

To differentiate itself from those rivals, Anthropic says, it’s focused on building more reliable models that aren’t affected by so-called “hallucinations,” which is the term used to describe models that generate fake answers. In addition, Anthropic has created guardrails to prevent toxic behavior such as racism, making its models “safer” than others, it claims. It says it specifically trains its models to generate “helpful, polite, respectful and thoughtful” responses and avoid offensive behavior.

Constellation Research Inc. analyst Holger Mueller said Anthropic’s launch of a paid plan shows how fast the generative AI market is growing up, with startups already looking for monetization models less than 12 months after the industry really kicked off. “It’s no surprise Anthropic wants to boost revenue as it’s known that creating, training and maintaining large language models is an expensive business,” he said. “Because the industry leader, OpenAI, is monetizing, that helps the rest of the industry generate organic revenue. Anthropic is following in its footsteps with Claude Pro. The downside of offering a paid model is that Anthropic gets less evaluating users, but the company seems to think it can strike the right balance between its free tier and paid offering.”

Anthropic said the Claude Pro subscription is available now, matching the $20-per-month cost of ChatGPT Plus.

Image: Anthropic