On Tuesday, Anthropic unveiled Claude, a large language model (LLM) that can generate text, write code and act as an AI assistant similar to ChatGPT. The model stems from key concerns about the safety of future artificial intelligence, and Anthropic trained it using a technique it calls “Constitutional AI.”
Two versions of the AI model, Claude and “Claude Instant,” are now available to a limited “early access” group and Anthropic’s commercial partners. Those with access can use Claudy either through Anthropic’s developer console chat interface or through the application programming interface (API). Through the API, developers can connect remotely to Anthropic’s servers and add Claude’s analytics and text completion capabilities to their applications.
Anthropic claims that Claude is “much less likely to produce harmful results, easier to chat with, and more controllable” than other AI chatbots, while maintaining a “high degree of reliability and predictability.” The company cites use cases such as search, summarization, collaborative writing, and coding. And, like ChatGPT’s API, Claude can change personality, tone, or behavior depending on usage preference.
To sell Claude, Anthropic charges usage per million characters of input and output. While OpenAI’s gpt-3.5-turbo AI model is $0.002 per 1000 characters (word fragment), Claude Instant is available for $0.42 per million characters as fast input and $1.45 per million characters for output. Claude-v1, the larger model, fetches $2.90 per million characters and $8.60 per million characters. While there is no standard conversion between tokens and characters, our back-of-the-envelope calculation is that OpenAI’s ChatGPT API is around $0.40 to $0.50 per million characters, so Claude is more expensive overall :
According to Anthropic, Claude has already been integrated into several products available through partners, such as DuckDuckGo’s DuckDuckGo Instant Summaries, part of Notion AI, and an AI chat app called Poe created by Quora. :
Dario Amodei, former vice president of research at OpenAI, and his sister Daniela founded Anthropic in 2021. how “AI security and research company” after disagreements over OpenAI’s commercial direction. Amody brought with him other OpenAI employees, such as Tom Brown, who led the engineering work on GPT-3. The GPT-3 revision later served as the core language model underlying ChatGPT. According to The Verge, Google has invested $300 million in Anthropic by the end of 2022 for 10 percent of the company.
Based on Google’s announced plans for Bard and the PaLM API, it seems unlikely that Google will rely on Anthropic for AI solutions in its products, but funding an OpenAI competitor could be in Google’s strategic interest. : For now, Anthropic plans to continue developing and improving Claude.
“We plan to introduce more updates in the coming weeks,” writes Anthropic. “As we develop these systems, we will continually work to make them more useful, honest and secure as we learn more from our security research and our deployments.”