OpenAI's newest Sol, in Rork

GPT-6.1 Sol is OpenAI's newest model in the GPT-6 series. OpenAI describes its capabilities as comparable to GPT-6 Astra, its most powerful model, with an unmatched mix of speed and affordability. It costs the same as GPT-6 Sol. It reads a 1M-token context, so it holds a large app in view at once.

How to use it

1. Open the model menu in the composer, on the home page or inside a project.
2. Pick GPT-6.1 Sol. It comes in three effort levels: Low, Medium, and High.
3. Describe your app or the change you want, and send.

GPT-6.1 Sol takes the place of GPT-6 Sol in the menu. If you picked GPT-6 Sol before, Rork now builds with GPT-6.1 Sol.

Availability

Pro and Max plans include GPT-6.1 Sol. On a free plan, the model menu shows the plan that unlocks it.

For a comparison of every model Rork offers, see the model menu in the composer. Availability and billing can change over time.

Anthropic's fastest Sonnet, in Rork

Claude Sonnet 5.5 is the newest Sonnet model from Anthropic. It has a sharp eye for UI design. It generates output over 30% faster than Sonnet 5. It reads a 1M-token context, so it holds a large app in view at once.

How to use it

1. Open the model menu in the composer, on the home page or inside a project.
2. Pick Sonnet 5.5. It comes in five effort levels: Low, Medium, High, Extra High, and Max.
3. Describe your app or the change you want, and send.

Sonnet 5.5 takes the place of Sonnet 5 in the menu. If you picked Sonnet 5 before, Rork now builds with Sonnet 5.5.

Availability

Pro and Max plans include Claude Sonnet 5.5. On a free plan, the model menu shows the plan that unlocks it.

For a comparison of every model Rork offers, see the model menu in the composer. Availability and billing can change over time.

Anthropic's new leader in agentic coding, in Rork

Claude Opus 5.5 is the first model in Anthropic's Claude 5.5 family. It is built for agentic coding, long-running tasks, and reasoning. It costs 40% less to run than Opus 5 and writes output over 30% faster. It reads a 1M-token context, so it holds a large app in view at once.

How to use it

1. Open the model menu in the composer, on the home page or inside a project.
2. Pick Opus 5.5. It comes in three effort levels: Low, Medium, and High.
3. Describe your app or the change you want, and send.

Rork builds with the model shown in the menu. You can switch models at any time from there.

Availability

Pro and Max plans include Claude Opus 5.5. On a free plan, the model menu shows the plan that unlocks it.

For a comparison of every model Rork offers, see the model menu in the composer. Availability and billing can change over time.

OpenAI's new GPT-6 model, in Rork

GPT-6 Sol is OpenAI's new GPT-6 model for complex coding and agentic work. It builds on the advances behind GPT-6 Astra in a faster, more affordable model. It costs half as much as GPT-5.6 Sol and makes about half the factual mistakes. It reads a 1M-token context, so it holds a large app in view at once.

How to use it

1. Open the model menu in the composer, on the home page or inside a project.
2. Pick GPT-6 Sol. It comes in three effort levels: Low, Medium, and High.
3. Describe your app or the change you want, and send.

Rork builds with the model shown in the menu. You can switch models at any time from there.

Availability

Pro and Max plans include GPT-6 Sol. On a free plan, the model menu shows the plan that unlocks it.

For a comparison of every model Rork offers, see the model menu in the composer. Availability and billing can change over time.

OpenAI's most capable model, in Rork

GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, and research. It reads a 1M-token context, so it holds a large app in view at once: many screens, a backend, payments.

How to use it

1. Open the model menu in the composer, on the home page or inside a project.
2. Pick GPT-6 Astra. It comes in three effort levels: Low, Medium, and High.
3. Describe your app or the change you want, and send.

Rork builds with the model shown in the menu. You can switch models at any time from there.

Availability

Pro and Max plans include GPT-6 Astra. On a free plan, the model menu shows the plan that unlocks it.

For a comparison of every model Rork offers, see the model menu in the composer. Availability and billing can change over time.

Rork Games

Say goodbye to game engines, Rork now gives you everything you need to build any game from scratch. Build 3D multiplayer shooters, 2D platformers, etc:

• Auto-generated high quality 3D models and animations
• Build games for the browser or mobile
• Multiplayer that just works
• Free in-game voice chat via Rork-hosted TURN servers

Check out our announcement post for more!

Rork Web

Rork can now create a web version of your iOS and Android apps that is accessible from a browser. Here are some cool things about it:

• Super fast one-click publish
• Custom domains!
• We built our own hosting layer to make it super fast

Check out our announcement post for more!

Bring your own Supabase

You can now connect your own existing Supabase project into your apps. This lets you reuse existing Supabase projects in Rork and use the Supabase dashboard.

If you want to transfer an existing Rork-managed Supabase project into your own Supabase account, you can prompt Rork to start the transfer process for you.

Top up credits

If you've consumed all your monthly credit usage, you can now "top up" credits and continue using Rork until your monthly credit limit resets.

We still recommend upgrading to higher-tier plan as the credits are discounted and will save you money on the long run.

To top up your credits, go to Profile > Credits > Top up credits

3D asset generation for every project

• 3D and asset generation now available on all plans (previously gated)
• Works across React Native, Kotlin, and Web projects
• Lighter, faster-loading models via texture compression
• Humanoid animations confirmed in parallel with mesh generation
• Unified 3D chat widget with explicit approval — no surprise auto-approve
• Faster 3D previews

Improvements

General

• Fixed an error with overly long messages sent to Rork
• Feature flags now evaluated server-side (also fixes ad/tracker-blocker breakage, e.g. Arc)
• Added a tooltip making it clearer why revert/undo is locked during a running message
• Support moved from the a little bubble to a Support menu item
• New mobile model picker — choose models on mobile web

3D

• Compressed GLB textures — models load lighter and faster

App Store

• CocoaPods prebuild failures now surfaced clearly

Performance

• 2-5x faster project page loads
• Rork starts 30–50% faster for large projects
• Faster streaming git sync

Fixes

General

• Typing in Chinese, Japanese, or Korean no longer sends your message mid-word
• Send button stays available while Rork is replying
• File saves safely blocked during project updates
• Chat no longer breaks when using your browser's auto-translate
• Back button no longer gets stuck closing dialogs

Previews

• Preview crashes now keep the full error details, so they get fixed faster
• Kotlin preview no longer flashes (install retries capped)

Hosting

• Legacy and Expo live URLs resolve to the right deployment
• Published sites recover instead of getting permanently stuck after a bad deploy

3D

• Cleaner Meshy animation GLBs (mesh and texture stripped)
• Animation preview playback fixed

Agent

• Inline images capped under provider limits — no more failed sends
• Unicode preserved in truncated context
• Toolkit SDK auto-install restored

App Store & mobile

• Fixed projects that could publish to the wrong App Store listing
• Fixed a freeze on the App Store publishing screen