Posted by 
While OpenAI is in the middle of an existential crisis, there’s a new chatbot update from Anthropic, the Google-backed AI startup founded by former OpenAI engineers who left over disagreements about the company’s increasingly commercial direction as its Microsoft partnership went on.
Anthropic has announced that the latest update of its chatbot, Claude 2.1, can digest up to 200,000 tokens at once for Pro tier users, which it says equals over 500 pages of material.
Our new model Claude 2.1 offers an industry-leading 200K token context window, a 2x decrease in hallucination rates, system prompts, tool use, and updated pricing.
Claude 2.1 is available over API in our Console, and is powering our https://t.co/uLbS2JNczH chat experience. p…
The company also says Claude will hallucinate — or lie — half as often as before, and it can do things like search the web or use a calculator through customizable tools. The chatbot also now supports custom, persistent instructions and has a new test window for trying out prompts.
The 200,000-token feature doubles what Claude could handle before, and it’s a significantly higher limit than the 32,000-token ceiling of the priciest version of GPT-4. Anthropic calls this an “industry first” and says it enables the chatbot to pore over uploads of entire codebases or works like the “Iliad.” The higher token count is exclusive to the company’s Pro plan.
Anthropic also brought Claude a little closer to ChatGPT with the new beta tool use feature. Users can now connect API tools, and Claude will pick the best one for the job depending on the context. That includes using a calculator or searching the web, and the company says users can ask for specific API calls using natural language.
Finally, the company said it’s updated its developer console with a test window for trying out new prompts and has added the ability to give Claude custom persistent instructions. Like with GPT-4, this means you can customize the chatbot to respond in specific ways or with specific personalities.
