Just days after the turmoil at OpenAI, which has forged doubt over the way forward for ChatGPT, Anthropic has launched an replace to its personal AI chatbot, Claude 2.1. This newest improve to Anthropic’s flagship mannequin not solely challenges OpenAI’s newest language fashions but additionally introduces distinctive options that set it aside.
What is Claude?
Claude, launched in March 2023 by Anthropic (established in 2021 by Dario and Daniela Amodei), works equally to ChatGPT. Claude may help with use circumstances together with summarization, search, inventive and collaborative writing, Q&A, coding, and extra.
What units Claude aside is its deal with security. Anthropic has made a aware effort to design Claude in a method that minimizes the danger of dangerous outputs, equivalent to bias, misinformation, and offensive language.
Claude powers two in style internet apps: Quora’s Poe and Notion.
What’s new in Claude 2.1?
The replace contains three main adjustments:
- Larger context window
- Reduced hallucination charges
- “API instrument use” for builders
Larger context window
Claude 2.1 has doubled the context window to 200K tokens, which is considerably bigger than the 128K token context window of OpenAI’s GPT-4. The 200K context window interprets to roughly 150,000 phrases, or over 500 pages of paperwork.
It means you possibly can add a whole ebook and discuss to it.
OpenAI’s GPT-4 presently has 128K context home windows, so which means Claude can deal with 62K extra tokens.
Note: The full 200K token capability stays unique to paying Claude Pro subscribers for now. Free customers will proceed to be restricted to Claude 2.0’s 100K tokens.
Reduced hallucination charges
Anthropic claims that Claude 2.1 has decreased the AI’s hallucinations and false statements by as much as 50%. This is a major enchancment, as hallucinations could be a main drawback with massive language fashions.
In an inside check carried out by Anthropic, Claude 2.1 was considerably extra prone to demur…
