Analysis
November 21, 2025

Claude 5 Wishlist: 5 Features We Need to See

From local execution to advanced artifacts, here is what the developer community is demanding from the next Claude.

Claude has become the go-to LLM for developers in 2024 and 2025. The "Artifacts" UI feature changed how we code. But to stay ahead of Copilot and Gemini, Claude 5 needs to evolve.

Here is our top 5 wishlist for the new model:

1. Stateful Memory (Project Awareness)

We don't just want a long context window. We want Claude 5 to "remember" our entire project structure permanently, without needing to re-upload files every session. Imagine a .cursorrules file but built natively into the model's brain.

2. Interactive Artifacts 2.0

Current Artifacts are great for React/HTML. But we want Claude 5 to deploy full-stack applications. Imagine asking for a backend, and Claude spins up a temporary container to run the Node.js server right in the chat.

3. "Thought Process" Control

OpenAI's o1 introduced hidden reasoning chains. We hope Claude 5 exposes this "inner monologue" but gives us control over it. Let us tell Claude: "Don't overthink this, just give me the regex."

4. Multimodal Output

Text is not enough. We want Claude 5 to generate UI mockups (Figma files), audio explanations, and perhaps even edit video clips directly.

5. Lower API Latency

For "Agentic" workflows where LLMs call other LLMs, Claude 3 Opus was sometimes too slow. Claude 5 needs to be as smart as Opus but as fast as Haiku.

What features do you want? Let us know on Twitter by tagging @AnthropicAI (and us!).