Anthropic just shipped Claude 3.5 Sonnet, its first model in the new 3.5 series. It’s faster and cheaper than Opus, and outperforms it on several key benchmarks, but the bigger story for builders is a new feature called Artifacts. This addition moves Claude from a simple conversational AI into a collaborative work environment, changing the core loop of how we build with these models. from chat loop to live workspace Historically, the workflow for using an LLM to code or create has been a turn-based conversation. You ask for a code snippet, you copy it, paste it into your local editor, run it, find an error, and then return to the chat to describe the problem. This cycle repeats, with the context split between your IDE and the chat window. The new Artifacts feature in the claude.ai interface changes this. When you ask for something like a React component or an SVG graphic, Claude now generates it in a dedicated panel next to the conversation. This panel is a live workspace where you can see the rendered output, edit the code, and iterate in real-time without leaving the browser. This creates a tighter feedback loop. Instead of copying and pasting, you can prompt for a change—"add a login feature"—and see the code and the rendered preview update in the same window. It’s a move toward making the AI an active participant in a shared workspace, not just a passive respondent in a chat. agentic coding gets a boost Beyond the UI, the model itself has significant improvements for developers. In an internal agentic coding evaluation, Claude 3.5 Sonnet solved 64% of problems, a marked improvement over the 38% solved by Claude 3 Opus. This evaluation tests the model's ability to fix bugs or add functionality to open-source codebases based on a natural language description. This suggests the model is more capable of complex reasoning, troubleshooting, and independent code execution when given the right tools. For teams working on migrating legacy applications or automating codebase maintenance, this is a meaningful step up. The model handles code translations and updates with more proficiency. Accessing the new model is straightforward. It's available through the Anthropic API, as well as on Amazon Bedrock and Google Cloud's Vertex AI. import anthropic client = anthropic.Anthropic( # defaults to os.environ.get("ANTHROPIC_API_KEY") api_key="my_api_key", ) message = client.messages.create( model="claude-3-5-sonnet-20240620", max_tokens=1024, messages=[ { "role": "user", "content": "Write a python function to check if a number is prime." } ] ).content[0].text print(message) The pricing is set at $3 per million input tokens and $15 per million output tokens, with a 200K token context window. This makes it significantly more cost-effective than Opus for tasks that require high intelligence but also demand speed, like orchestrating multi-step agentic workflows. what this means for builders The release of Claude 3.5 Sonnet and the Artifacts feature isn't just another incremental model bump. It's a signal of where the developer experience is heading. The friction of context-switching between a chatbot and an IDE is being actively designed out of the process. This evolution from conversational partner to collaborative tool is the key takeaway. For builders, it means we can start designing workflows that assume the AI is in the editor with us, not just in another window. This has implications for prototyping, debugging, and even how we onboard new developers to a complex codebase. The model is becoming an on-demand teammate. Sources https://www.anthropic.com/news/claude-3-5-sonnet