Introducing Orchids Agent-1

Today we're excited to introduce Orchids Agent-1: 4x smarter, 3x cheaper.

Agent-1 is already live for all users, with no migration, setup, or toggle required on your end. If you're building in Orchids today, you're already using it.

We're also making GPT 5.4 available to all users starting today.

Try it now at orchids.app.

A smarter default agent, automatically

The goal with Agent-1 was simple: make Orchids meaningfully better without adding more complexity for users. Instead of asking you to switch versions, update settings, or learn a new workflow, we replaced the underlying default agent experience directly.

That means better reasoning, better tool use, and better output quality in the same product flow you already know. Agent-1 is designed to be more reliable across the kinds of tasks people use Orchids for every day, from planning and scaffolding to debugging and iterative product work.

Most upgrades in AI tooling come with some amount of friction. New model picker, new pricing math, new workflow caveats. We wanted this one to feel different. The improvement is already in place, and the experience should just feel faster, sharper, and more useful.

4x smarter, 3x cheaper

The headline for Agent-1 is straightforward because the impact is straightforward too.

• 4x smarter
• 3x cheaper

Those gains matter because the best AI product experiences are not only about raw capability. They are about capability per dollar and capability per prompt. Smarter systems reduce retries, reduce cleanup work, and reduce the need to constantly restate context. Lower cost means users can push further on real projects without the experience feeling wasteful.

In practice, Agent-1 is built to help you get to strong first drafts faster, recover from dead ends more easily, and spend less time fighting the model when you are trying to ship.

GPT 5.4 is now available to all users

Alongside Agent-1, GPT 5.4 is now available across Orchids for all users.

That gives builders another top-tier model option inside the platform, whether they are refining app behavior, iterating on product logic, or pushing on harder implementation tasks. Model quality matters most when the work stops being boilerplate and starts involving tradeoffs, judgment, and multi-step execution, so we want the strongest options available inside Orchids by default.

We believe the best AI product should make model improvements immediately useful instead of burying them behind complicated configuration.

Built for real product work

Agent-1 is not just about sounding better in a chat reply. It is about improving the actual work of building.

That includes things like:

• stronger task understanding
• better iteration quality
• more useful responses on complex requests
• lower cost when working through longer projects

As AI tools mature, the gap between an impressive demo and a useful product becomes more obvious. We care about the latter. The point of Orchids is not to show a clever generation once. It is to help people actually build and ship.

Availability

Agent-1 is available now for all users, and there is nothing you need to do to turn it on.

One important note: Agent-1 is currently available only when using GPT or Claude models.

If you're already building with Orchids, the upgrade is live. If you haven't tried it yet, now is a good time to jump in and see how far you can get with the new default experience.

What's next

Agent-1 is another step toward making Orchids the fastest way to go from idea to working software. We are continuing to improve model quality, reliability, and the end-to-end product experience so that more of the building process feels automatic and more of your energy can go into the work that actually matters.

We're excited for you to try it.