تحديث في 19/11/2026
• A new llama.cpp version, with better support for models on your device.
• New standard and thinking models. Ones you've already installed stay on the device.
• Thinking, follow-up replies, graphics acceleration, and Regenerate are more reliable.
• Your API model lists update without an app update.
• A more polished interface, with steadier steps and file handling in Code Lab.
• Standard, Pro, and Max for fast cloud models, plus high-quality voice input.
