Today, we’re releasing Anaconda Desktop in public beta, a single application for AI development that brings model discovery, local inference, and conda environment management together for the first time.
Anaconda Desktop is free to download. You can find a model and build with it locally, all from one place. It also runs on the same infrastructure the Python community has trusted for over a decade.
Built for where you are now
For millions of people, Anaconda Navigator has been the front door. It’s where students launch their first notebook and feel something unlock. It’s where researchers spin up reproducible environments for work that matters. It’s where engineers build with Python packages and get their work off the ground.
But the world of data science and development today looks nothing like it did a decade ago. Large language models have moved from the periphery to the center of what teams are building, and they brought their own stack with them. Now, you have to manage a server and an API layer alongside your package manager and environment setup. None of them natively work together, so you have to cobble together solutions that do.
Anaconda Desktop solves that problem. It takes what Navigator started and extends it to the full AI development workflow. It’s built to own the whole thing.
One place for models, inference, and environments
Anaconda Desktop puts local LLMs in the same place as your environments and your packages. For beta, we’re releasing:
- Full conda environment management. Create, clone, and manage conda environments through a visual interface. Launch your preferred IDE, like JupyterLab or Spyder, with the right environment already active.
- A curated AI model catalog. Browse and filter models by your hardware, quantization, task type, license, or publisher. Compare model scoring benchmark performance across reasoning, code, and function calling before you commit to a download. Not sure where to start? An AI-powered recommendation tool helps you find the right model for your use case.
- Local inference built in. Download a model and run it in chat or as a local API server your applications can connect to—no separate inference tool to configure.
If you’ve been cobbling together a local AI stack, you probably have a model hub here, an inference tool there, and a separate API layer for your application. Desktop replaces that with a single surface. There’s no need to switch tools, so you won’t waste time debugging dependency conflicts between them.
Already an Anaconda user?
Everything you relied on in Anaconda Navigator is still here: creating and managing conda environments, installing packages, launching Jupyter Notebooks, and more. But now, local AI capabilities sit right alongside them.
Transitioning from Navigator to Desktop is not a migration. Your environments, packages, and custom Jupyter builds will carry over automatically.
- Download Anaconda Desktop, available for Windows, Mac, or Linux.
- Install it on your machine.
- Open it. Your existing conda environments will be right there waiting for you.
Desktop finds what’s already on your machine and displays it for you. You don’t need to recreate anything or run any commands. If something looks off or your environments aren’t showing up, the most common cause is multiple conda installations on your machine.
If you have features you want to see or things that should work differently, tell us on our feedback board. This product is built in conversation with the community, and your input matters.
We’ll continue to support Anaconda Navigator through the end of 2026 for existing users, but we encourage you to try out the new Desktop experience.
What’s coming next
The public beta is just the start. Here’s where we’re taking it.
Later this summer, you’ll be able to deploy and manage multiple inference endpoints, locally or in the cloud, from Anaconda Desktop, bridging the gap between development and production. We’re also expanding Anaconda MCP to give AI agents direct, governed access to the conda ecosystem.
We’re working out how models, MCP servers, agents, and skills all fit into the AI software development lifecycle, and what it looks like when Anaconda sits at the center of that stack. The goal is to keep accelerating developer velocity, practically and securely.
Try it out
Anaconda Desktop is free. You’ll need an Anaconda account to get started, which also gives you access to our curated model catalog.
If you’re heading to PyCon US in Long Beach in May, come by the Anaconda booth #313. We’ll be demoing Desktop in person, and I’ll be there and happy to dig into questions.