Support for Opus 4.7.
Translation settings now include an option to disable model thinking.

MLX local models support Gemma 4

Support for the latest models

Support Foundation Models.

Add Nvidia provider.

Add MiniMax provider.

MLX (local model) supports the Qwen3.5 model and supports enabling and disabling thinking mode.
Supports the latest online models.

fix bugs and performance improvement.

Supports the latest models. Fixed multiple bugs.

Fixed the issue of duplicate tool calls.
Added Claude Compatible providers to support Coding Plans from various vendors.
Added a "Use Responses API" toggle in the OpenAI Compatible provider settings.

Supports Responses API, Cloud's GPT series models support web search.
Thinking, search, and tool selection will no longer be reset when switching conversations.
Support for the latest models.
Model menu displays model price multiplier.
Other UI optimizations and bug fixes.

Support latest models.

Gemini and Anthropic providers support  search.
Gemini and Claude models in Cloud support search.
Thinking mode changed: When the model supports thinking, a mode selection will be displayed to choose between quick response or deeper thinking.

Support latest models. Fixed an issue when using tools in conversation.

Fixed incorrect effort values passed during API calls for certain models.

Supports the latest models. Sidebar supports opening and closing with gestures.

Optimize message sending experience and automatically align messages to the top.
Fix the issue where replies are interrupted when switching to the background.
Fix some UI issues.

Support latest models.

Fixed UI issues

Support 5.2 models.
Fixed a critical UI unresponsiveness bug on iOS 17.

Fixed occasional crashes when displaying code blocks.
Fixed parameter errors not supported by Poe API.
Added support for DeepSeek 3.2.
Added a read-aloud reply toggle switch in the top-right corner of the conversation.

Fix issues where responses could be interrupted or not displayed when using certain third-party providers due to non-standard format returns.
Optimize keyboard extension experience.
Optimize performance.

Support the latest model.
Third-party service providers will no longer display the default model list after adding custom models.
Fixed the issue where text-to-speech usage was not displayed.
Fixed UI details.

New feature: uses drawer-style navigation.
Settings allow you to change whether the default chat is a new chat or a casual chat.

Support latest model.
Fixed temperature error in translator.