Launch Week
New From Anaconda
New tools and capabilities across the full Anaconda platform — from local AI development to enterprise environment governance.
Secure by Default
Security bolted on at the end of a workflow is damage control, not governance. Anaconda’s Environment Services Manager embeds vulnerability visibility, policy enforcement, and compliance directly into your team’s process, so security travels with your environments automatically.
GA / Cloud Only
Environment Services Manager
- Know your exposure fast: instantly see affected environments, owners, and package versions for any vulnerability.
- Automatic capture, zero workflow changes: every conda environment is logged from first login with no disruption.
- Built for audits: every environment change creates a tamper-proof history searchable by user, date, or package.
Trusted AI Workflows
AI initiatives stall when experimentation can’t cross the gap to production. We close that gap by introducing the tools and integrations that make AI workflows reproducible, governed, and observable from your local environment to cloud to production deployment.
Anaconda + Outerbounds
Anaconda Outerbounds
- One workflow, any destination: Build once, deploy consistently, reproducible from laptop to cloud to production
- Experiment, deploy, and observe on one governed platform across the entire AI lifecycle
- No gaps, no surprises with orchestration, GPU compute, and observability on Anaconda’s trusted foundation.
GA / MCP
Anaconda MCP
- One integration, full conda awareness: connect MCP-compatible AI tools to your real environments, channels, and packages.
- No more pip guesswork: agents use conda, read .condarc, and resolve dependencies against your actual system state.
- Enterprise policies travel with the agent: channel restrictions and governance stay enforced automatically.
- Configure once, benefit continuously: new MCP servers and Skills appear automatically with no reconfiguration.
Developer Velocity
New tools to help AI builders move faster. One command configures your full Python toolchain, one app handles model discovery and environment management, and a new governed channel delivers 6,000+ packages. Spend less time fighting infrastructure and more time delivering.
Open Beta / Desktop
Anaconda Desktop
- Skip the tool-switching: one app for model discovery, local inference, and environment management.
- Pick the right model before you download: filter by hardware, task, and license; compare benchmarks first.
- Your existing Anaconda environments and packages carry over automatically — zero re-setup.
Open Beta / CLI
Anaconda CLI
- Install once: a single executable sets up Miniconda and configures conda + pixi with secure defaults
- Connect AI coding assistants to your Anaconda toolchain through the Anaconda MCP Server
- Build and deploy directly from the same CLI
- Access configured channels with no manual setup or configuration drift
GA Release / Anaconda Core
AI Without the Assembly
- PyTorch releases now targeted within 30 days of upstream — eliminating the 5+ month wait that forced teams to mix conda-forge or pip into their environments.
- PyTorch GPU support expanded to Windows CUDA (2.7+) and Linux ARM64/aarch64 (2.9+) — one channel now covers workstations, DGX clusters, and standard Linux-64 deployments.
- 19 production AI packages now on main — covering the full workflow from RAG and agent orchestration to fine-tuning, inference, and model monitoring, all professionally integrated and tested together.
Open Beta / Main-X
New main-X channel
- Close the package gap: main-x adds 6,000+ Anaconda-built conda packages, with thousands more coming soon.
- Keep teams secure: Packages are built on Anaconda’s validated infrastructure with malware and typosquat protection.
- Zero workflow disruption: main-x works alongside the existing main channel with standard conda tooling — no new plugins, no migration, no retraining required.
The Foundation Your AI Projects Have Been Missing
See what’s possible when your entire AI stack is built on a trusted foundation.