Anthropic’s latest model, Claude 3.5 Sonnet, has dethroned OpenAI’s GPT-4o and secured the top spot in the Coding Arena and Hard Prompts Arena. It was also placed second on the Overall leaderboard. “Claude is so much better at coding than ChatGPT. Night and day difference (sic),” wrote a user on X.
The new Sonnet has surpassed Claude Opus at five times the lower cost and is competitive with frontier models like GPT-4o and Gemini 1.5 Pro across the board.
However, the best is yet to come. Anthropic plans to release Claude 3.5 Haiku and Claude 3.5 Opus later this year, along with developing new features like Memory for personalised user experiences.
Play with Artifacts
The standout feature in Claude 3.5 Sonnet that has captured everyone’s attention is the Artifacts tool. It appears as a separate window in the corner, assisting users in visualising their tasks.
When users request Claude to generate content such as code snippets, text documents, or website designs, the Artifacts feature displays these creations in a dedicated window within their conversation interface. Currently, OpenAI’s ChatGPT does not boast of any such feature, nor has it announced anything similar.
This setup creates a dynamic workspace, allowing users to view, edit, and enhance Claude’s outputs in real-time. It seamlessly integrates AI generated content into their projects and workflows.
Since its release, many users have been experimenting with the model to create games, websites, functional AI sound effects, and simulations.
“Wow! Claude 3.5 Sonnet with Artifacts will revolutionise learning! I asked Claude to create an animation simulating projectile motion, and the result blew me away! Imagine generating custom animations and illustrations for any topic in minutes. Goodbye, static textbooks! @brilliantorg should be scared! ” wrote a user on X.
Similarly, AI investor Allie K Miller used Claude 3.5 Sonnet to create an interactive educational tool to teach her ‘daughter’ how AI is used to communicate with animals.
Anthropic has also introduced the ‘Projects’ feature in Claude where users can now organise chats with Claude into shareable projects. Each project includes a 200K context window, equivalent to a 500-page book, allowing users to incorporate all relevant documents, code, and insights to enhance Claude’s effectiveness.
Moreover, users can set custom instructions within each project to further tailor Claude’s responses.
Codes Like Flash
“Claude 3.5 Sonnet is better at writing code than a typical computer science grad,” said former Stability AI chief Emad Mostaque.
Source: X
When equipped with the necessary tools, Claude 3.5 Sonnet can autonomously write, edit, and execute code, showcasing advanced reasoning and troubleshooting abilities. It excels at code translations, making it especially useful for updating legacy applications and migrating codebases.
“Been using Claude 3.5 for coding day-to-day and wow, I think this may happen sooner than people imagine. Software is dead. And we have killed him,” posted another user on X.
Furthermore, when combined with Artifacts, Claude 3.5 Sonnet simplifies developers’ tasks by providing real-time visibility into their app development process. This capability makes Claude an ideal pair programmer, enhancing collaboration and efficiency.
Wake Up OpenAI!
Lately, OpenAI has faced mounting pressure from its competitors. The company announced the GPT-4o model just before Google I/O, but its voice capabilities have yet to be released. “It’s been a month and a half and there is no new voice or vision functionality available, openai shipped a blog post,” quipped a user on X.
OpenAI has announced that the voice feature will be made available to all Plus users this fall. “We had planned to start rolling this out in alpha to a small group of ChatGPT Plus users in late June, but need one more month to reach our bar to launch,” the company said.
The company is improving the model’s ability to detect and refuse certain content. “We’re also working on improving the user experience and preparing our infrastructure to scale to millions while maintaining real-time responses.”
OpenAI will start the alpha with a small group of users to gather feedback and expand based on what they learn. “We are planning for all Plus users to have access in the fall.”
“Just a year ago, it was unthinkable that any other model would even remotely approach GPT’s lead. Today, Sonnet-3.5 (not even Anthropic’s biggest Opus model) is already slightly above and Llama-3-400B is around the corner,” posted another user on X.
“I never thought I would have anything on par with the OpenAI models, but Anthropic is killing it with every new model without any drama, focused on APIs and developers instead of giving startup-killing vibes and hubris. And when we hit the wall, Anthropic could be a winner,” said KissanAI founder Pratik Desai.
Meanwhile, Amazon is developing a ChatGPT killer app, internally codenamed Metis. This project uses retrieval-augmented generation to provide up-to-date information and automate tasks.
The same thing happened with OpenAI’s video generation model Sora. “Just 4 months ago, Sora blew everyone’s mind and seemed so out of reach. Today, we have at least 4-5 clones of Sora at 70-80% quality, such as Kling, Luma, and Runway. The clones wouldn’t have rallied without OpenAI’s first move,” posted NVIDIA AI researcher Jim Fan.
One can only hope that OpenAI will keep up with their promises and deliver the new models soon.
