Claude 3.5 Sonnet is here but is it better than GPT-4o?

Anthropic revealed its latest model Claude 3.5 Sonnet, a faster and better version of its predecessor. Compared to its previous version, Claude 3 Opus, this model is a big step up. It’s quicker, understands humor better, and can even read your handwriting. Obviously that is not it and the company revealed some important features alongside the announcement.

There is no doubt that Anthropic is one of the biggest competitors to Microsoft and OpenAI when it comes to LLMs and artificial intelligence in general. They even have a huge fanbase that thinks they are way better in many aspects, yet these people prefer Anthropic’s technology over others. Clause 3.5 Sonnet is the latest “miracle” that the company revealed and it quickly caught the interest of industry leaders.

What is Claude 3.5 Sonnet?

Claude 3.5 Sonnet is Anthropic’s newest model. It’s designed to talk to users more smoothly, faster, and cheaper, making powerful AI tools more available to everyone. You can use it for free or pay for more features if you are a pro user.

Claude 3.5 Sonnet comes with impressive features

Claude is great at many things. It’s quicker at completing tasks, can handle many languages, and understands complex problems better. One major improvement is how it looks at images. It can now understand pictures, charts, and even handwritten notes very well. This makes jobs like reading text from photos much easier.

Claude 3.5 Sonnet comes with some exciting updates. It’s faster and costs less, which makes it attractive to both regular users and developers. This model is just the start. Anthropic plans to update other models like Haiku and Opus soon. For now, Sonnet is a standout because of its speed and price.

One exciting new feature is called ‘Artifacts.’ This feature lets the AI show its work directly. For example, if you ask it to make a piece of code or design a website, it can display the results right beside the chat. This makes Claude more than just a chatbot. It’s becoming a key tool for managing projects and automating workflows.

Claude 3.5 Sonnet vs. GPT-4o

Claude 3.5 Sonnet has some notable upgrades over its earlier version, Claude 3 Opus. It is twice as fast and much cheaper, making it both quick and cost-effective. It also handles more information at once compared to GPT-4o, thanks to its larger context window of 200K versus 128K in GPT-4o. This feature is especially useful for dealing with long documents or detailed discussions.

Vellum AI conducted an analysis on these models to see how they perform in specific tasks like extracting data from legal contracts, classifying customer support tickets, and solving verbal reasoning problems like math riddles.

• Data Extraction: Both Claude 3.5 and GPT-4o were tested on their ability to pull out information from legal contracts. They both identified 60-80% of the data correctly. However, neither model stood out, suggesting that both need better prompting techniques for high accuracy in such complex tasks.
• Customer Ticket Classification: In this task, the models determined whether customer support tickets were resolved. Claude 3.5 Sonnet achieved a 72% accuracy rate, slightly better than GPT-4o’s 65%. But GPT-4o was more precise, with an 86.21% precision rate, important for making sure tickets are classified correctly. Claude’s precision was a bit lower at 85%.
• Verbal Reasoning: This test involved questions on math riddles and analogies. GPT-4o performed better, especially in tasks that required detailed calculations or identifying opposites. It had a 69% accuracy rate, whereas Claude struggled with numbers and dates, only achieving 44% accuracy. However, Claude did well on analogy questions.

These tests by Vellum AI show that each model has strengths in different areas. Claude 3.5 Sonnet is great for tasks that need fast processing and are sensitive to cost. However, for jobs that require high accuracy and detailed reasoning, GPT-4o is usually better.

Claude AI pricing