Release Notes 🚀✨

What’s New ✨
• Refactored model-alias management across providers 🔄
– Removed many legacy aliases (e.g. “gpt-4.1-mini”, “phi-4”, “grok-3-mini”)
– Unified alias lookup via get_alias → simpler, more reliable
• PollinationsAI improvements 🌸
– Streamlined get_models / alias handling
– Cleaned up text/vision model lists
– Dropped outdated aliases (e.g. deepseek-v3) in favor of direct model IDs
• Ollama enhancements 🦙
– Support for local models + proxy usage
– Automatic API key loading
– Fixed infinite loop in streaming generator
• Azure provider cleanup ☁️
– Removed dead extra_body parameter
– Simplified streaming logic

New & Updated Providers 🆕
• EasyChat
– Added “share auth” flow & HTTP status checks 🔐
– Only streaming mode now (no more sync responses)
• OpenRouterFree 🎛️
– Brand-new free-tier support via OpenRouter API
• ApiAirforce ✈️
– New provider for api.airforce.gg

LMArena suite improvements 🎯
• Added model-map sync at startup
• Fixed model IDs, reset logic & list population
• Beta endpoint updates in LMArenaBeta.py

Search providers & deprecations 🔍
• Moved DuckDuckGo / Free2GPT into deprecated/ folder
• SearXNG revamp:
– Renamed default_url → url
– Prompt formatting via format_media_prompt
– Language parameter & empty-result handling
– Proxy support & debug logging

Broken / disabled in this release 🚫
• DeepSeekAPI (needs_auth) set to working = False
• You.com streaming search disabled (working = False)

Bugfixes & Misc 🐛
• Removed leftover debug prints
• Guard added in EasyChat to error if no args found
• Cleaned up imports & class attributes across providers

Thank you to everyone who contributed! 🙏
— The g4f Team