Anthropic’s newest Claude chatbot beats OpenAI’s GPT-4o in some benchmarks

Anthropic rolled out its newest AI language model on Thursday, Claude 3.5 Sonnet. The updated chatbot outperforms the company’s previous top-tier model, Claude 3 Opus, while working at twice the speed. Claude users (including those on free accounts) can check it out beginning today. Sonnet, which tends to be Anthropic’s most balanced model, is the first release in the Claude 3.5 family. The company says Claude 3.5 Haiku (the fastest in each generation) and Claude 3.5 Opus (the most powerful) will arrive later this year. (Those models will stay on version 3 in the meantime.) The Sonnet update comes only a few months after the arrival of the Claude 3 family, showcasing the breakneck speed AI companies are working to spit out their latest and greatest. Anthropic Anthropic claims Claude 3.5 Sonnet marks a step forward in understanding nuance, humor and complicated prompts, and it can write in a more natural tone. Benchmarks (above) show the new model breaking industry records for graduate-level reasoning, undergraduate-level knowledge and coding proficiency. It beats OpenAI’s GPT-4o on many of the benchmarks Anthropic published. However, the latest Claude, ChatGPT, Gemini and Llama models tend to score within a few percentage points of each other on most tests, underscoring the tight competition. The company claims Claude 3.5 Sonnet is also better at interpreting visual input than Claude 3.0 Opus. Anthropic says the new model can “accurately transcribe text from imperfect images,” a skill it hopes will attract customers in retail, logistics and financial services who need to grok data from charts, graphs and other visual cues.  Claude’s update also brings a new workspace the company calls Artifacts (above). When you prompt the chatbot to generate content like code, text documents or web designs, a dedicated window appears to the right of the chat. From there, you can prompt Claude to make changes, and it will keep the Artifacts window updated with its latest output. The company sees Artifacts as a first step towards making Claude a space for broader team collaboration. “In the near future, teams — and eventually entire organizations — will be able to securely centralize their knowledge, documents, and ongoing work in one shared space, with Claude serving as an on-demand teammate,” the company wrote in a press release. Claude 3.5 Sonnet is available now for anyone with an account to try on its website, as well as in the Claude iOS app. (On both of those platforms, Claude Pro and Team subscribers get higher token counts.) You can also access it through the Anthropic API, Amazon Bedrock and Google Cloud’s Vertex AI. It costs $3 per million input tokens and $15 per million output tokens — the same as the previous model.This article originally appeared on Engadget at https://www.engadget.com/anthropics-newest-claude-chatbot-beats-openais-gpt-4o-in-some-benchmarks-170135962.html?src=rss

Jun 20, 2024 - 23:30
 0
Anthropic’s newest Claude chatbot beats OpenAI’s GPT-4o in some benchmarks

Anthropic rolled out its newest AI language model on Thursday, Claude 3.5 Sonnet. The updated chatbot outperforms the company’s previous top-tier model, Claude 3 Opus, while working at twice the speed. Claude users (including those on free accounts) can check it out beginning today.

Sonnet, which tends to be Anthropic’s most balanced model, is the first release in the Claude 3.5 family. The company says Claude 3.5 Haiku (the fastest in each generation) and Claude 3.5 Opus (the most powerful) will arrive later this year. (Those models will stay on version 3 in the meantime.) The Sonnet update comes only a few months after the arrival of the Claude 3 family, showcasing the breakneck speed AI companies are working to spit out their latest and greatest.

Chart showing benchmarks comparisons between recent AI chatbot models: Claude 3.5 Sonnet, Claude 3 Opus, GPT-4o, Gemini 1.5 Pro and Llama-400b.
Anthropic

Anthropic claims Claude 3.5 Sonnet marks a step forward in understanding nuance, humor and complicated prompts, and it can write in a more natural tone. Benchmarks (above) show the new model breaking industry records for graduate-level reasoning, undergraduate-level knowledge and coding proficiency. It beats OpenAI’s GPT-4o on many of the benchmarks Anthropic published. However, the latest Claude, ChatGPT, Gemini and Llama models tend to score within a few percentage points of each other on most tests, underscoring the tight competition.

The company claims Claude 3.5 Sonnet is also better at interpreting visual input than Claude 3.0 Opus. Anthropic says the new model can “accurately transcribe text from imperfect images,” a skill it hopes will attract customers in retail, logistics and financial services who need to grok data from charts, graphs and other visual cues. 

Claude’s update also brings a new workspace the company calls Artifacts (above). When you prompt the chatbot to generate content like code, text documents or web designs, a dedicated window appears to the right of the chat. From there, you can prompt Claude to make changes, and it will keep the Artifacts window updated with its latest output.

The company sees Artifacts as a first step towards making Claude a space for broader team collaboration. “In the near future, teams — and eventually entire organizations — will be able to securely centralize their knowledge, documents, and ongoing work in one shared space, with Claude serving as an on-demand teammate,” the company wrote in a press release.

Claude 3.5 Sonnet is available now for anyone with an account to try on its website, as well as in the Claude iOS app. (On both of those platforms, Claude Pro and Team subscribers get higher token counts.) You can also access it through the Anthropic API, Amazon Bedrock and Google Cloud’s Vertex AI. It costs $3 per million input tokens and $15 per million output tokens — the same as the previous model.This article originally appeared on Engadget at https://www.engadget.com/anthropics-newest-claude-chatbot-beats-openais-gpt-4o-in-some-benchmarks-170135962.html?src=rss

What's Your Reaction?

like

dislike

love

funny

angry

sad

wow

Viral News Code whisperer by profession, narrative alchemist by passion. With 6 years of tech expertise under my belt, I bring a unique blend of logic and imagination to ViralNews360. Expect everything from tech explainers that melt your brain (but not your circuits) to heartwarming tales that tug at your heartstrings. Come on in, the virtual coffee's always brewing!