g4f v0.4.7.5

What's Changed

• Add new media selection in UI by @hlohaus in #2760
• Add HuggingFace provider provider
• Auto refresh Google Gemini cookies
• Add sources to search results

Full Changelog: 0.4.7.4...0.4.7.5

g4f v0.4.7.4

What's Changed

• Fix url download function by @hlohaus in #2744
• Update interference-api.md by @whoa-mi in #2749
• AI Provider and Model Updates: Adding New, Removing Deprecated, and Enhancing Functionality by @kqlio67 in #2739
• Fix model and provider in chat completion response by @hlohaus in #2753

New Contributors

• @whoa-mi made their first contribution in #2749

Full Changelog: 0.4.7.3...0.4.7.4

g4f v0.4.7.3

Release Notes

This release brings several improvements and bug fixes to g4f, focusing on performance, user interface enhancements, and stability.

Key Changes:

• Performance Improvements:
  • Implemented WakeLook functionality to potentially improve background task performance.
  • Disabled token counting to boost overall performance.
• User Interface Enhancements:
  • Added functionality to export and import settings directly within the UI.
  • Made various styling improvements to enhance the user experience.
• Bug Fixes:
  • Resolved bugs related to curl_cffi updates, ensuring better compatibility and stability.

Pull Request:

• #2743: (16 Feb) by @hlohaus - Includes all the changes listed above.

Full Changelog:

• v0.4.7.2...0.4.7.3

Note: This release focuses on refining existing features and addressing reported issues. Users are encouraged to update to benefit from the performance and stability improvements.

g4f v0.4.7.2

What's Changed

• Use gradio api in flux dev by @hlohaus in #2738

Full Changelog: 0.4.7.1...0.4.7.2

g4f v0.4.7.1

What's Changed

• Update space demo by @hlohaus in #2737

Full Changelog: 0.4.7.0...0.4.7.1

g4f v0.4.7.0

What's Changed

• Improve tools support in OpenaiTemplate and GeminiPro by @hlohaus in #2734
• Update models in DDG, PerplexityLabs, Gemini
• Fix issues with curl_cffi in new versions
• Show only free providers by default
• Add mixtral_small_24b model from DDG

PydanticAI Integration with G4F Client

Quick Setup

1. Patch G4F to Use PydanticAI Models

In order to use PydanticAI models with G4F, you need to apply the necessary patch to the client. This can be done by importing apply_patch from g4f.tools.pydantic_ai.

from g4f.tools.pydantic_ai import apply_patch
apply_patch()  

2. Create a Simple Agent

Now you are ready to create a simple agent that can interact with the LLM. The agent is initialized with a model, and you can also define a system prompt. Here's an example where a basic agent is created with the model g4f:Gemini:Gemini and a simple system prompt:

from pydantic_ai import Agent

# Define the agent
agent = Agent(
    'g4f:Gemini:Gemini', # g4f:provider:model_name or g4f:model_name
    system_prompt='Be concise, reply with one sentence.',
)

Read More: PydanticAI Integration with G4F Client

Full Changelog: 0.4.6.2...0.4.7.0

g4f v0.4.6.2

What's Changed

• Small fixes and updates around various providers by @Arken1xs in #2717
• Fix errors on read buckets by @hlohaus in #2725

New Contributors

• @Arken1xs made their first contribution in #2717

Full Changelog: 0.4.6.1...0.4.6.2

g4f v0.4.6.1

What's Changed

• Fix styling on mobile by @hlohaus in #2708

Full Changelog: 0.4.6.0...0.4.6.1

g4f v0.4.6.0

Release Notes

We're excited to announce the release of g4f 0.4.6.0! This update brings several new features, improvements, and bug fixes aimed at enhancing both functionality and user experience. Below is a summary of the key changes:

---

New Features

• Vision Models for HuggingFace Provider
  Expanded the HuggingFace provider to support vision models, opening up advanced image processing capabilities alongside traditional text models.
  [github.com]

• Premium Model Support in Blackbox Provider
  Introduced premium model support with enhanced validation in the Blackbox provider. This improvement ensures a smoother experience with robust premium authentication workflows.
  [github.com]

• Enhanced Unicode Image Handling
  Improved image processing in the copy_images module by adding better support for Unicode. This change enhances compatibility when dealing with files containing special characters.
  [github.com]

• Improved Mobile Interface
  Upgraded the client interface with enhanced mobile responsiveness and refined sidebar interactions, ensuring a more seamless user experience on mobile devices.
  [github.com]

---

Improvements and Bug Fixes

• Demo Mode Stability
  Resolved issues that were impacting demo mode functionality, resulting in a more stable and reliable demonstration environment.
  [github.com]

• API Endpoint Corrections
  Fixed provider-specific API endpoint errors to guarantee smoother data exchange and better overall connectivity with various services.
  [github.com]

• CablyAI Provider Refactor
  Reorganized the CablyAI provider by moving it into a dedicated needs_auth directory. This refactor clarifies its authentication requirements and simplifies future maintenance.
  [github.com]

• PollinationsAI Alias Cleanup
  Removed duplicate model aliases in the PollinationsAI provider to prevent conflicts and ensure accurate model selection during runtime.
  [github.com]

• Pipeline Tag Optimization for HuggingFace API
  Implemented caching for the pipeline tag and corrected its checking mechanism. These changes optimize performance by reducing redundant API calls and ensuring tag accuracy.
  [github.com]

• CLI and Initialization Enhancements
  Updated the command-line interface and initialization routines, streamlining user interactions and improving overall stability.
  [github.com]

---

Miscellaneous

• General Codebase Improvements
  Additional minor fixes and refactors have been merged into the codebase, enhancing code quality and laying the groundwork for future updates.
  [github.com]

---

We thank all the contributors for their hard work and valuable feedback. Enjoy the new features and improvements in g4f 0.4.6.0!
For more details, please check the full commit comparison.

New Contributors

• @mikebilly made their first contribution in #2706

g4f v0.4.5.5

Release Notes

New Features

• Introduced zerogpu_uuid: A new feature added to the demo, allowing for better identification and tracking in GPU-less environments.
• FluxDev Integration: The G4F provider now supports FluxDev, enhancing the variety of models available for users.

Improvements

• Enhanced Copy Images API: Improvements made to the Copy Images API for better performance and user experience.
• Updated Webmanifest: The webmanifest has been revised to include the latest features and optimizations.

Bug Fixes

• Resolved "Show New Window" Button Issue: Fixed a bug that prevented the "Show New Window" button from functioning correctly.
• Exception Handling in Generate Function: Added checks to handle exceptions in the generate function, improving overall stability.

Commits Overview

• Feb 5, 2025:

  • Added zerogpu_uuid to demo.
  • Implemented exception checks in the generate function.

• Feb 6, 2025:

  • Integrated FluxDev into the G4F provider.
  • Updated webmanifest and added missing imports.

• Feb 7, 2025:

  • Fixed the "Show New Window" button functionality.
  • Merged pull request #2688 and added new models while enhancing provider functionality.

What's Changed

• Add zerogpu_uuid to demo by @hlohaus in #2688
• Adding New Models and Enhancing Provider Functionality by @kqlio67 in #2689

For a complete comparison of changes, please refer to the commit history here [github.com].