Releases: DahnM20/ai-flow
v0.8.1
Version v0.8.1 - Standalone Windows Desktop Release
Download the release here : https://ai-flow.net/release/
If you're not a Windows user, you can run the app using docker-compose. Or use the Cloud version at app.ai-flow.net.
New Features:
- Enhanced Web Extractor node with additional fields for customized data extraction
- Improved Web Extractor speed
- Faster document loader for PDF files
- Introduction of GPT-4o-mini
- Resolution of processor launcher concurrency issues
- Smoother initial loading experience
- Various UI enhancements and bug fixes
Visit docs.ai-flow.net for configuring the file upload, if needed.
Please note that when using the Stable Diffusion/Text-to-speech/File Upload nodes locally, files are stored on your machine. Therefore, they cannot be used as input for other nodes such as the Replicate Node or GPT Vision. To make it work, you will need to set a S3 bucket to upload files automatically. You can find more info on docs.ai-flow.net
v0.7.2
Version v0.7.2 - Standalone Windows Desktop Release
Download the release here : https://ai-flow.net/release
If you're not a Windows user, you can run the app using docker-compose. Or use the Cloud version at app.ai-flow.net.
New Features
- Full StabilityAI API now available. (To fully benefit from it, you'll need to set up file upload.)
- Claude 3 now available.
- Data Splitter updated to support a new "Manual" mode.
- Layout View updated with a grid layout and the ability to add headings and text.
- New Help action available for each node, with links to related content on the wiki/blog.
- Visit docs.ai-flow.net to configure the file upload node if needed.
- Various UI adjustments and bug fixes
Please note that when using the Stable Diffusion/Text-to-speech/File Upload nodes locally, files are stored on your machine. Therefore, they cannot be used as input for other nodes such as the Replicate Node or GPT Vision. To make it work, you will need to set a S3 bucket to upload files automatically. You can find more info on docs.ai-flow.net
v0.7.0
Version v0.7.0 - Standalone Windows Desktop Release
Download the release here : https://ai-flow.net/release
If you're not a Windows user, you can run the app using docker-compose. Or use the Cloud version at app.ai-flow.net.
New Features
- New resizable Display Node
- GPT4o now supported
- Output text is now adjustable
- Nodes can have custom names
- Flows can have custom names
- Color can be chosen for each node
- 3D file formats .obj and .glb are now supported
- Models with multiple outputs are now handled
- New configuration tab to enable/disable nodes
- Layout View items are now on a transparent background (still in progress)
- Be sure to visit docs.ai-flow.net to make the file upload node work if needed
- Various UI adjustments and bug fixes
Please note that when using the Stable Diffusion/Text-to-speech/File Upload nodes locally, files are stored on your machine. Therefore, they cannot be used as input for other nodes such as the Replicate Node or GPT Vision. To make it work, you will need to set a S3 bucket to upload files automatically. You can find more info on docs.ai-flow.net
v0.6.4
Version v0.6.4 - Standalone Windows Desktop Release
Download the release here : https://ai-flow.net/release
If you're not a Windows user, you can run the app using docker-compose. Or use the Cloud version at app.ai-flow.net.
New Features
- New Stable Diffusion 3 node
- New Text-to-Speech node
- New Document-to-text node
- You can now add your own nodes easily in the app, see our Guide
- Be sure to see docs.ai-flow.net to make the file upload node work if needed
- Various UI Adjustments and bug fixes
Please note that when using the Stable Diffusion/Text-to-speech/File Upload nodes locally, files are stored on your machine. Therefore, they cannot be used as input for other nodes such as the Replicate Node or GPT Vision. To make it work, you will need to set a S3 bucket to upload files automatically. You can find more info on docs.ai-flow.net
v0.6.3
Version v0.6.3 - Standalone Windows Desktop Release
Download the release here : https://ai-flow.net/release
If you're not a Windows user, you can run the app using docker-compose. Or use the Cloud version at app.ai-flow.net.
New Features
- New "Open in Sidepane" action available for every node.
- Nodes parameters and output can be displayed in a new sidepane view.
- Sliders are more precise.
- Outputs can be hidden in the "Outputs" view of the sidepane.
- Interactions with node actions have been improved.
- Various UI Adjustments and bug fixes
Please note that when using the Stable Diffusion Node locally, images are stored on your machine. Therefore, they cannot be used as input for other nodes such as the Replicate Node or GPT Vision. An alternative is to use the DALL-E or any text-to-image model that is accessible via the Replicate Node. It's also possible to set a S3 bucket to upload automatically the Stable Diffusion files.
v0.6.2
Version v0.6.2 - Standalone Windows Desktop Release
Download the release here : https://ai-flow.net/release
If you're not a Windows user, you can run the app using docker-compose.
New Features:
- Quick Actions on Node Click: Includes options to Remove, Duplicate, and Clear Output.
- New File Upload Node: Functional with a configured S3 bucket.
- Templates Menu: Introduces a selection of basic flow examples.
- Saving Flow as a Template: Now available as a feature.
- Enhanced Node Interaction: Added scrolling capability when the cursor is on a node output.
Additional Improvements:
- Various bug fixes and UI improvements.
Server Version Release (.exe)
Available here : https://ai-flow.net/release
Please note that when using the Stable Diffusion Node locally, images are stored on your machine. Therefore, they cannot be used as input for other nodes such as the Replicate Node or GPT Vision. An alternative is to use the DALL-E or any text-to-image model that is accessible via the Replicate Node. It's also possible to use a S3 bucket to upload automatically the Stable Diffusion files.
v0.6.1
Version v0.6.1 - Windows Executable Release
Download the release here : https://ai-flow.net/release
-
GPT answers are now streaming.
-
New Transition Node to help organize your workflow.
-
Introduced a new button to remove edges.
-
Improved error handling for nodes missing a field; nodes in error are now highlighted.
-
Bug fixes for specific issues with the Replicate Node.
Please note that when using the Stable Diffusion Node locally, images are stored on your machine. Therefore, they cannot be used as input for other nodes such as the Replicate Node or GPT Vision. An alternative is to use the DALL-E or any text-to-image model that is accessible via the Replicate Node. It's also possible to use a S3 bucket to upload the Stable Diffusion files.
v0.6.0
Version v0.6.0 - Windows Executable Release
Download the release here : https://ai-flow.net/release
-
NEW Replicate Node : Experience a diverse range of open-source models with the new Replicate Node, featuring popular models Llama, Mistral, Stable Video Diffusion, Music-gen, custom SDXL, and many more.
-
Edge Type Selection : New top-right UI button for selecting and using different Edge Types.
-
Additional Information : Due to the extensive range of models available via Replicate, some may not be supported yet. Also, as the Replicate API costs are based on processing time and hardware type, pricing may vary greatly between models.
To use the Replicate Node, you'll need to set REPLICATE_API_KEY in your environnement, or update the .env file in server/.env.
Please note that when using the Stable Diffusion Node locally, images are stored on your machine. Therefore, they cannot be used as input for other nodes such as the Replicate Node or GPT Vision. An alternative is to use the DALL-E or any text-to-image model that is accessible via the Replicate Node.
v0.5.2
Version v0.5.2 - Windows Executable Release
This update is focused on optimizing the overall user experience with key enhancements:
- Concurrent Node Execution: Nodes can now run as soon as they have obtained the necessary data to be launched.
- URLLoader Optimization: To further streamline performance, the URLLoader has been moved to a background thread. Other processing tasks are no longer hindered by URL loading operations.
v0.5.0
Version v0.5.0 - Windows Executable Release
🌟 New features 🌟
-
Customizable Layout View: Dive into efficiency with the new Layout view feature. Tailor your workflow by displaying only the sections you need, such as input and output nodes, perfect for large and complex flows.
-
Versatile Merge Node: Experience flexibility with the innovative Merge Node. It comes in two flavors:
- Simple Merge: Seamlessly combine two inputs into one.
- GPT-Prompt Merge: Craft GPT prompts by merging two inputs with placeholders. For instance, you can use
Improve ${input1} based on ${input2}to specify how the inputs should merge.
-
UI Improvements: Enjoy a smoother user interface with our latest enhancements that make navigation and operation more intuitive and user-friendly.
Description
This release, as usual, provides a user-friendly Windows executable for those who like a direct approach. Just download, and run server.exe - Note that you may need to run server.exe as an administrator for it to function correctly.
When you have the server up, simply go to localhost:5000.
For other users, the project can quickly be set up on Linux and/or through Docker.
Please refer to the README for installation and usage instructions.
Feedback is the essence of improvement. Got ideas or suggestions? Don’t hesitate to share ! 😄