Anthropic, the AI startup founded by former executives from OpenAI, has released an updated version of its text-generating model, Claude Instant. The new model, called Claude Instant 1.2, incorporates the strengths of Anthropic’s flagship model, Claude 2, and shows significant improvements in areas such as math, coding, reasoning, and safety.
In internal testing, Claude Instant 1.2 outperformed its previous version, scoring 58.7% on a coding benchmark compared to 52.8% for Claude Instant 1.1. It also achieved an 86.7% accuracy on a set of math questions, compared to 80.9% for Claude Instant 1.1.
According to Anthropic, Claude Instant 1.2 generates longer and more structured responses, follows formatting instructions better, and also shows improvements in quote extraction, multilingual capabilities, and question answering. The updated model is also less likely to generate incorrect or nonsensical text (hallucination) and more resistant to jailbreaking attempts.
Claude Instant 1.2 features a context window that is the same size as Claude 2’s, which is 100,000 tokens. The context window refers to the text the model considers before generating additional text. This allows the model to analyze approximately 75,000 words, similar to the length of “The Great Gatsby.”
Anthropic’s ultimate goal is to create a next-generation algorithm for AI self-teaching, which can be used to build virtual assistants capable of answering emails, performing research, and generating various forms of content. However, Claude Instant is not the algorithm itself but rather a competitor to other entry-level text-generating AI systems offered by companies like OpenAI, Cohere, and AI21 Labs.
Anthropic currently has thousands of customers and partners, including Quora, which provides access to Claude and Claude Instant through its generative AI app Poe. Claude also powers DuckDuckGo’s DuckAssist tool, which provides direct answers to search queries, and is integrated into Notion AI, an AI writing assistant for the Notion workspace.
Since its launch in 2021, Anthropic has raised $1.45 billion in funding, but the company estimates it will need $5 billion over the next two years to realize its vision for a fully functional chatbot.