GPT Engineer logo

Changelog

New updates and improvements.

Better Version Management

You may often find yourself in a situation where something breaks, and you need to revert to a previous stable version that was working correctly. However, up until now, finding that specific version has been a difficult and time-consuming process.

We’ve made it easier for you to quickly locate past edits by adding new descriptive labels to each version. We’ve also made the preview capability more visible, allowing you to view the version before choosing to revert to it.

In addition, we’ve made it super easy to quickly access the last 8 edits. This should make managing your project’s evolution smoother and more intuitive.

Version management feature in action.

12x Speed Improvements in Large Projects

Large projects that accumulated a significant number of edits used to be very slow, as they would consume a lot of memory in our browsers. We’ve made significant improvements to how messages are rendered.

Rather than loading all messages at once, they are now rendered on demand as you scroll through the chat history. This change resolves the performance slowdowns some users experienced, making the editor more responsive.

Smaller Fixes and Improvements

  • We have lowered the threshold for when we rely on RAG, which has reduced errors by 25% in smaller projects.
  • GPT Engineer is now less eager to generate code immediately and more open to having exploratory conversations, allowing you to effectively plan or discuss things before diving into code changes.
  • We fixed a bug where GPT Engineer would fail to install dependencies.
  • We have open-sourced GPT Engineer’s documentation! If you notice an issue or want to share best practices, you can now contribute directly via our docs or by heading to our GitHub repo.

We’re always happy to hear any feedback you have. To reach us, you can:

Branch Switching for More Control

By default, GPT Engineer commits changes directly to the main branch. Now, you have the flexibility to choose which branch GPT Engineer commits to. You can configure this under Projects Settings.

This feature is for those who prefer a PR-focused workflow or that have CI/CD pipelines that automatically deploy on changes to the main branch. It can also serve those that want to manage experimental work separately, ensuring that only finalized updates make their way to your main branch.

Branch switching feature in action.

Hot Projects Tab for Inspiration

Looking for inspiration or curious to see what others are building with GPT Engineer? We’ve introduced a Hot Projects tab that showcases trending public projects so that you can see what’s generating buzz.

We’ll continue to improve the algorithm that selects these projects, so expect even better recommendations in the coming weeks. You can check out the current hot projects here by visiting the project overview and selecting the Hot tab.

Hot projects feature in action.

Improved Network Error Detection

GPT Engineer can now detect and surface network errors. If an issue arises with API requests or external services, you’ll be notified with an error, enabling you to quickly address and resolve them.

Network error feature in action.

Better Context Management for Large Projects

More and more of our users move beyond simple prototypes and build larger, more complex applications with GPT Engineer. When projects get big, language models have a harder time attending to the right context, and in some cases GPT Engineer simply wouldn’t work.

To solve this, we’ve introduced Retrieval-Augmented Generation (RAG), which identifies and retrieves the relevant code for each request. This enhancement allows you to continue building significantly larger projects without compromising the performance of GPT Engineer.

Enhanced Handling of Laziness

Language models have a tendency to be lazy, using placeholders like “// rest of the code” instead of writing the actual code. To address this, we’ve implemented automatic detection of these placeholders and now seamlessly fill in the blanks to ensure the code is complete and functional.

We’re always happy to hear any feedback you have. To reach us, you can:

Shareable Profiles

In our last update, we introduced user profiles. You can now share your profile URL (e.g., https://gptengineer.app/@fabian) to showcase your creations with others!

Other users can also visit your profile by clicking on your username in the latest project view.

user profiles.

Add Custom Knowledge

You can now add custom knowledge to your project that will provide GPT Engineer with any additional context it might need.

This could include documentation on a specific library or any other important guidelines. By doing so, you ensure that GPT Engineer has all the relevant information to deliver the best results.

You will find the “Manage knowledge” option in the project settings.

add custom knowledge

”Edit with GPT Engineer” Badge Removed for PRO Users

By popular request, we’ve removed the “Edit with GPT Engineer” badge from published deploys for PRO users.

edit with gpt engineer badge

If you want this change to apply to existing deploys, simply redeploy your project.

Smaller Bugs / Improvements

  • Improved Prompt: The prompt has been updated to reduce laziness that can introduce unwanted placeholders.
  • Supabase Integration Fixes: We’ve resolved issues where the Supabase integration could get stuck, adding reliability.

Let Us Know What You Think

We’re always eager to hear your feedback. To reach us, you can:

We made #1 Product of the day & week. It took us by surprise. We scrambled to add an ‘alpha’ label to GPT Engineer, and we heard loud and clear signals we were onto something people wanted.

But that was just the beginning. Since that surprise hunt, we’ve personally onboarded hundreds of users from our waitlist, and relentlessly iterated based on it.

This wasn’t just about polishing rough edges; we’ve been experimenting with the very core of GPT Engineer. We tried multiple agentic AI architectures, combinations of LLM models, and abstractions to interface with models, edit code, and provide the iterative feedback loop needed to build real apps in real time.

The goal? To strike a balance between speed and reliability, give seasoned developers a tool to turbocharge productivity, integrate with workflows they love, while lowering the barrier to entry for anyone to create software.

Graduating from alpha to beta

We’re really proud of where GPT Engineer is today. In our latest iterations, we’ve found that founders, CTOs, product engineers, and maintainers of internal tools are able to build front-end & ship software faster than ever before. With that, we’re graduating from ‘alpha’ to ‘beta’.

The move to beta is more than a label change. Generating a new front-end takes on average 15 seconds, to go from prompt to prototyped. Edits take on average seconds — not yet WYSIWYG — but quick enough to stay in flow while iterating towards a viable product. Errors can be auto debugged by AI. Deploying takes a single click. We’ve made huge strides in both reliability, and capabilities.

In part, we’ve been able to achieve this by being opinionated in GPT Engineer’s tech stack. Front-end is built React, Tailwind & Vite. GPT Engineer can integrate directly with Supabase & OpenAPI specification backends, but for now we recommend using it to build front-end, and taking over in your IDE from there (everything syncs with GitHub, two-way).

Join the waitlist & shape the roadmap

The beta still has a waitlist, but if you’re a founder, product engineer, CTO, or internal tool builder — get in touch for immediate access, and be part of shaping the next phase of GPT Engineer to your needs.

We’re committed to a future where anyone can build software, and we believe this is just the beginning.

We’re excited to have you all along for the journey.

Let us know what you think

We’re always happy to hear any feedback you have. To reach us, you can:

Faster Edits

We’ve significantly enhanced the editing experience, making it faster and more responsive, especially for smaller, granular edits.

View Changes in Realtime

With every edit, a loading indicator will now appear in the chat, which you can click to view real-time changes to your files. This feature offers immediate feedback, helping you track modifications as they happen.

Real-time code change viewing.

Subscriptions

We’ve updated our pricing. Free users now get 10 free edits per day, while PRO users receive 100 daily edits included in their $19/month subscription. For more details, visit our subscription page. Additionally, PRO users can now make private projects.

Tip: If you’re a PRO user and want to set all your new projects to private by default, you can do so in your profile under Settings.

Pricing overview.

User Profiles

We’ve introduced user profiles, allowing you to set your username, location, and other personal details. Claim your unique username before someone else does!

User profiles.

Renaming Projects

You can now easily rename your project by navigating to Settings -> Rename Project. Please note that this will not change the name of your GitHub repository.

Small Fixes

  • All GitHub projects will now default to being private.

Let Us Know What You Think

We’re always eager to hear your feedback. You can reach us by:

A new Lovable

We’ve just upgraded our website, including adding this blog. Expect to hear more from the team on upcoming features, deep dives, and behind the scenes sneak peeks.

Redesigned Task Interface

We’ve revamped the way we show cards to unify progress indicators and edit options into a single task card. This new design simplifies tracking your ongoing tasks.

New task cards in GPT Engineer.

Stop Button Added

You can now halt any active task running in the background by clicking the stop button located in the input area. This feature aims to reduce stress when you need to stop an agent that is off track or using an approach you disagree with.

Stop button in GPT Engineer.

Shadcn as the Default UI Component Library

Shadcn is now the default design system for new projects. However, if you prefer the look and feel of Chakra UI, you can still select it through the advanced settings.

Shadcn/UI support in GPT Engineer.

Let us know what you think

We’re always happy to hear any feedback you have. To reach us, you can:

UX Revamp

We have completely revamped the design of the homepage and editor. More updates are on the way!

UX changes to GPT Engineer.

New agent flow

We have been building and testing a more “agentic” flow that should feel more conversational, flexible, and in the long term also more resilient that our “classic” flow. Prefer the classic method? No problem—you can easily opt out in Settings before starting a new project. Note: Once you create a project using a given modality, you cannot change the modality.

Select agent or classic in GPT Engineer.

Page selector

You can now preview any page directly by selecting it from our newly added page selector dropdown.

Select any page in GPT Engineer.

Images in prompt

Guide GPT Engineer’s responses more effectively by attaching images to your prompts. Try it with a mockup or color theme today!

Note: For best results, make sure you reference the image in the prompt so that the agent uses it in the desired manner.

Upload images to prompt GPT Engineer.

## NPM Packages

Now, adding advanced features like graphs, drag & drop, and maps is easier. GPT Engineer automatically installs the NPM packages your will need for this.

Contacting us

We’re always happy to hear any feedback you have. To reach us, you can:

You can now use GPT Engineer to build a frontend that will work with an existing backend services, by specifying your API’s OpenAPI specification URL in the project settings. Once you do this, GPT Engineer will become aware of your backend’s capabilities and take this into consideration when iterating on the frontend.

Specify an OpenAPI backend specification for GPT Engineer.

Expanded support to new Tech Stacks

Have you ever wanted to create a project using Tailwind? We’ve now introduced support for additional tech-stacks beyond React + Chakra UI, including:

  • Tailwind with React
  • Simple HTML
  • Tailwind & Daisy UI

Choose your stack in the Advanced Settings before initiating a new project generation.

Choose your tech stack with GPT Engineer.

New foundation model integration

Accelerate your project builds with our new integration of the Claude 3 Haiku model — expect gains in speed and efficiency.

New foundation model integration in GPT Engineer.

Contacting us

We’re always happy to hear any feedback you have. To reach us, you can:

Fixes

  • Added a custom 404 page for instances where a project is not found.
  • Added more troublehoot tips to the modal shown for improve errors.