0.4.3.6

Release Notes for Version 0.4.3.5

Date: January 30, 2025

Commits and Updates:

1. Added Features:

   • Introduction of the Janus_Pro_7B provider to enhance functionality and user options.

2. Improvements:

   • Added reasoning capabilities in PerplexityLabs and updated the model list for better accuracy and efficiency.
   • Added cookies_to_dict for Copilot to streamline the handling of cookies.

3. Bug Fixes:

   • Retry Provider script updated to ensure more reliable retry mechanisms.
   • Raised MissingAuthError exception for invalid access tokens, ensuring better error handling and user feedback.
   • Passed missing DeepSeekAPI import to resolve issues caused by absent dependencies.

4. Pull Requests:

   • Merged pull request #2643 from hlohaus/demo to incorporate the latest changes and fixes.

For a detailed view of the changes, please refer to the GitHub comparison page

g4f v0.4.3.5

Release Notes for GPT4Free Version v0.4.3.5

Release Date: January 28, 2025

Key Updates:

1. Addressed Issue #2628:

   • Refactor parsing logic for improved handling of reasoning content
   • Commit: 2591781

2. Added AsyncAuthedProvider in Copilot:

   • Enhanced the Copilot integration with asynchronous authenticated providers.
   • Commit: 9524c3f

3. Added Model qwen-2.5-1m-demo to models.py:

   • Included a new model in the models.py file for demonstration purposes.
   • Commit: 17d2f05

4. Fixed Unittests and Set Strategy for Rate Limiter:

   • Addressed issues with unittests and implemented a new strategy for rate limiting.
   • Commit: 2c89583

5. Added PreviewResponse Type to Response Types:

   • Introduced a new type for preview responses.
   • Commit: d4f736e

6. Support for Options Like max_depth in downloads.json:

   • Added support for various options, including max_depth, in the downloads.json file.
   • Commit: efacc15

New Contributors

• @SreejiBR made their first contribution in #2630

Full Changelog: 0.4.3.4...0.4.3.5

g4f v0.4.3.4

What's Changed

• Demo by @hlohaus in #2629
• Add Feature provider in demo
• Support default provider in DDG
• Read api_key from config file

Full Changelog: 0.4.3.3...0.4.3.4

g4f v0.4.3.3

What's Changed

• Fix report errors by @hlohaus in #2624

Full Changelog: 0.4.3.2...0.4.3.3

g4fv 0.4.3.2

Release Notes for v0.4.3.2 (G4F)

What's New:

• Fix for file uploads in demo mode – Ensures files can be uploaded seamlessly during demo use.
  (PR: #2621)

• Updates for memory with mem0 – Memory handling has been improved with the introduction of mem0.

• Fix for asyncio import in nodriver function – Fixed an issue where asyncio wasn't being imported properly, improving functionality.

• Provider-specific API endpoints – Support added for API endpoints that are specific to individual providers.
  (PR: #2621)

• Open settings in UI – You can now open settings in the UI at /chat/settings.

• Qwen_Qwen_2_5M_Demo provider added – A new provider, Qwen_Qwen_2_5M, is now available.
  (PR: #2621)

• Support for <think> token in OpenAI Template – The <think> token is now supported in OpenAI templates for advanced interactions.
  (PR: #2621)

• Improved mobile UI – Fixed image selection issues on mobile devices and added support for boxed inline MathJax.

• Check request limit only in demo mode via API – Request limit checking is now only done when using the demo API.

• UI Improvements:

  • Stop recognition in the UI when pressing Enter.
  • Fixed rate limit handling for the "Ping" request in the GUI.
  • Error reporting has been integrated into the UI.
  • Support for custom providers in demo mode.

Full Changelog:

• See the full changelog for all the updates and fixes in this release:
  v0.4.3.1...v0.4.3.2 Changelog

---

This update brings a variety of improvements to the user interface and backend functionality, addressing important issues and expanding support for new features.

g4f v0.4.3.1

What's Changed

• Updates for Blackbox and DeepInfraChat providers new models by @kqlio67 in #2608
• Fix some issues in the demo by @hlohaus in #2619

Full Changelog: 0.4.3.0...0.4.3.1

g4f v0.4.3.0

Release Notes for Version 0.4.2.7

Release Date: January 26, 2025

Highlights

This release introduces significant enhancements, optimizations, and new features to improve functionality, robustness, and user experience. Special thanks to H Lohaus for their valuable contributions.

---

Key Updates and Features

Enhancements to docker-compose

• Port Update: Added 1337:8080 to docker-compose-slim.yml and docker-compose.yml for enabling new functionalities.
• Improved compatibility and setup flexibility across all running applications.

Provider Script Improvements

• Refactored and optimized providers such as CablyAI, Cloudflare, and DeepInfraChat to streamline inheritance and logic.
• Introduced use_nodriver attribute for enhanced customization in multiple provider classes.
• Added a format_prompt_qwen2 function to manage new prompt formats in HuggingFace.

HuggingFaceAPI Enhancements

• Better token handling with added max_tokens and calculate_length functions for accurate token counting and input management.
• Enhanced robustness through improved handling of messages exceeding length limits.
• Streamlined model properties for better management.

Demo Mode Features

• New demo.html file, complete with supporting CSS and JS, to enable and showcase demo functionalities.
• Updated backend to support demo mode, including functions like create_app_with_gui_and_debug and create_app_with_demo_and_debug.
• Integrated user-journey constraints for better demo tracking and usage monitoring.

Rate Limiting

• Integrated flask_limiter for API rate limiting in demo mode, improving UI stability under public access.

Error Handling Enhancements

• Improved edge case management by:
  • Ignoring encoding errors in file uploads and web_search.
  • Enhancing fault tolerance in processes.

Code Structure and Robustness

• Numerous refactorings and fixes to improve maintainability, performance, and clarity across the codebase.

---

Recommendations for Testing

1. Demo Mode: Test the new UI demo mode extensively, covering both frontend and backend functionalities.
2. Token Handling: Validate HuggingFaceAPI with various models to ensure the new max_tokens limit functions as expected in production.
3. Rate Limiting: Verify the behavior of flask_limiter under varying traffic conditions in demo mode.
4. Error Handling: Ensure encoding error handling manages issues gracefully without bypassing critical workflows.

---

What's Changed

• Update provider capabilities and model support by @kqlio67 in #2600
• Update README.md by @Pashok111 in #2612
• Increase max token in HuggingfaceAPI by @hlohaus in #2614
• Fix issue reading provider model list by @hlohaus in #2618

New Contributors

• @Pashok111 made their first contribution in #2612

Thank you for using version 0.4.2.7! 🚀

Full Changelog: 0.4.2.6...0.4.3.0

g4f v0.4.2.6

What's Changed

• Update bulding and publishing docker image for arm64 by @hlohaus in #2605
• Fix async_generator can't be used in 'await' expression on Windows

Full Changelog: 0.4.2.5...0.4.2.6

g4f v0.4.2.5

Release Date: January 25, 2025

New Features

• OpenaiTemplate Provider: Added support for the OpenaiTemplate provider.
• MiniMax Providers: Introduced MiniMax providers and included them in the model list.
• HailuoAI Provider: Added support for the HailuoAI provider.
• Default Vision Model: Set a default vision model for better consistency.

Improvements

• Docker Image Update: Updated the base image of the main Docker image for improved performance and compatibility.
• Model Updates: Added MiniMax to models.py to reflect the latest supported models.
• Lock File for nodriver: Added a lock file to ensure proper handling of the nodriver setup.

Bug Fixes

• Browser Management: Ensured the browser is properly stopped after all calls by adding a finally block.
• Unit Test Fixes: Addressed issues in unit tests for more reliable testing.

Removals

• Deprecated Webdriver Providers: Removed outdated or unused webdriver providers to streamline functionality.

by @hlohaus

Full Changelog: 0.4.2.4...0.4.2.5

g4f v0.4.2.4

What's Changed

• Add "Selecting a Provider" Documentation by @hlohaus in #2599
• Fix generate Images with OpenaiChat
• Add "flux" as alias in HuggingSpace providers
• Choice a random space provider in HuggingSpace provider
• Update requirements list in pypi packages
• Fix label of CablyAI and DeepInfraChat provider

Full Changelog: 0.4.2.3...0.4.2.4