- Published on
Launch Week Day 1: Leap 2.0
- Authors
- Name
- Marcus Kohlberg
We launched Leap three months ago and we've been blown away by your interest and support for what we're building. What we've achieved together over the past few months is incredible:
Today we're launching Leap 2.0, a major upgrade that handles larger projects more reliably, and without the cost explosion you see with other AI tools.
Most AI coding tools work fine for small demos. But as projects grow, they start to break down; making more mistakes, slowing down, and getting expensive fast. Leap 2.0 changes that. It now has tools so it can reason about big projects, makes much fewer errors, and remains dramatically cheaper as the codebase grows.
Leap 2.0 comes packed with new features, let's jump in!
| Feature | Leap 1.0 | Leap 2.0 | |
|---|---|---|---|
| Model Power | Claude 3.5 | Claude 4.5 | → |
| Project Size | Small to medium apps | Enterprise-scale systems | → |
| Web Browsing | Not available | Full web access | → |
| Image Generation | Not available | Nano Banana integration | → |
| Build Process | Manual error fixing | Automatic detection & fixes | → |
| Reasoning Modes | Standard | Think, Think hard, Think harder, Ultrathink | → |
Smarter agent, stronger reasoning
Leap 2.0 is powered by Claude 4.5, giving it a much deeper understanding of structure, intent, and dependencies across large codebases.
You can now guide its reasoning, and push Leap to solve even the hardest problems, using simple keywords:
It's not just generating code anymore. It's reasoning through architecture, debugging, and design.
Web browsing and docs reading
Leap can now browse the web and read documentation while it works. That means it can pull in up-to-date docs for third-party integrations and get things right the first time.
Visual creativity built in
Leap now integrates Nano Banana for generating and editing images directly in your repo. You can now create logos, visual assets, or marketing images instantly, without leaving your workflow.
Build and fix automatically
Leap runs builds as it generates code, spots issues, and fixes them immediately. You end up with working code, and way less errors. This makes it possible to move from idea to production in hours, not days.
Human-in-the-loop when it matters
When Leap needs a judgment call, like when making design tradeoffs, it doesn’t guess. It pauses, asks for your input, and continues with your direction. You stay in control of your project’s decisions while Leap handles the heavy lifting.
Built for production, not just demos
Leap deploys to your own AWS or GCP environment using clean, auditable infrastructure. You keep full control over your code, data, and deployment. No lock-in, no black boxes.
And because Leap 2.0's usage costs scales gently as your app grows, you can take on much larger projects without seeing costs skyrocket. When you deploy to your own cloud, you pay AWS/GCP directly, keeping control of your destiny. That's a huge shift from many other AI builders, which lock you in to using their proprietary Backend as a Service platforms.
Try Leap 2.0
Leap 2.0 is available today: Start building for free
Stay tuned
This is just the beginning! Stay tuned for more exciting announcements throughout Launch Week as we drop 4 more powerful features to make your AI driven development better.
Ready to try it out? Sign up and start building today!
– The Leap Team
Questions? Join us on Discord or email us at hello@leap.new.