What's Changed

Features

• Support Dark Mode on Android, iOS and Mac (Following system settings).
• Support adding richer document formats（'json', 'py', 'ts', 'tsx', 'js', 'kt', 'java', 'swift', 'c', 'm', 'h', 'sh', 'cpp', 'rs', 'go', 'class', 'cs', 'php', 'rb', 'dart', 'sql', 'css', 'xml', 'yaml'）as plain text in user prompt.
• Support adding documents for Ollama models and models compatible with OpenAI format.
• Support thinking mode for Claude sonnet 4 and Claude opus 4.
• Support New Nova Sonic Voice ID Lupe and Carlos in Spanish.

Other Changes

• Fix the duplicate rendering issue caused by audio waveform updates.
• Fix crash issue for add voice system prompt.

Full Changelog: 2.3.0...2.4.0

What's Changed

Features

• Support speech to speech by Amazon Nova Sonic on iPhone, iPad and Mac 🚀
  • Support Custom Voice Conversation System Prompt.
  • Support Selecting Voice ID on Settings page.
  • Support Barge In, also you can disable in system prompt.
  • Support Echo Cancellation, you can talk directly to the device without wearing headphones.
  • Support Voice Waveform to display realtime volume level changes.
• Support display Latency and token response speed metrics
• Change question text UI into a bubble format, update side bar background color on mac
• Support use proxy for OpenAI Compatible models
• Add Support for GPT-4.1 series model

Other Changes

• Fix OpenAI Compatible Reasoning model Reasoning text do not display issue
• Fix Bedrock model chunk parse error by add converse/v3 api
• Fix the jittery scrolling issue when content doesn't fill a full screen

Full Changelog: 2.2.0...2.3.0

Upgrade Notice !!!

Starting from this version, we begin to use the /converse/v3 API, which requires:

1. Upgrade your CloudFormation Stack, please refer upgrade-cloudformation
2. Upgrade your API, please refer: upgrade-api

What's Changed

Features

• Support OpenAI API Compatible Models
• Support quick switch text model by click bottom right model icon
• Support manual interruption of scrolling for streaming
• Add regenerate icon for regenerate any history AI response
• Add edit icon for auto select all raw markdown text
• Add an eye icon to show API keys in Settings page
• Support dollar($ and $$) format latex display
• Support loading for wait first streaming response
• Support loading when open long text history session
• Support tap message title to copy all response messages

Other Changes

• Add notice for using Chinese prompt to generate images when nova-lite is not enabled
• Add notice for DeepSeek models not support multi-modal

Full Changelog: 2.1.0...2.2.0

What's Changed

Features

• Add support for DeepSeek-R1 model on Amazon Bedrock

Full Changelog: 2.0.0...2.1.0

What's Changed

Features

• Support Claude 3.7 Sonnet with thinking mode.
• By optimizing the caching strategy, significantly improve Markdown rendering performance 🚀
• Added /converse/v2 API for better parse the reasoning text.
• Increase the lambda timeout to 15 minutes.
• Update the UI display of the reasoning content, and support the rendering of reasoning in Markdown format.
• Fixed known issues.

Full Changelog: 1.10.0...2.0.0

Note: Starting from this version, we begin to use the /converse/v2 API, which requires upgrading your API, please refer: upgrade-api

What's Changed

Features

• Support for Ollama text and vision models
• Support for DeepSeek models: DeepSeek-V3 and DeepSeek-R1
• Support for OpenAI models: GPT-4o and GPT-4o Mini

Full Changelog: 1.9.0...1.10.0

Note: Using OpenAI API Proxy feature needs to upgrade your API, please refer: upgrade-api

What's Changed

Features

• Support fully customizable System Prompt Assistant 🎉
• Support Permanent Sidebar display mode for Mac
• Change default Image model to Stable Diffusion 3.5 Large

Others

• Upgrade react-navigation to 7.x
• Support native stack with drawer layout.

Full Changelog: 1.8.0...1.9.0

Note: Using System Prompt feature needs to upgrade your API, please refer: upgrade-api

What's Changed

Features

• Support LaTeX display for Amazon Nova.
• Change default text model to Nova Pro and change default translate model to Nova Lite when generate images.

Others

• Support GitHub action for release new version.

Full Changelog: 1.7.0...1.8.0

What's Changed

Features

1. Support for Using Natural Language to Generate and Edit Images with Nova Canvas

• Enable generation of image variations.
• Facilitate color-guided image generation.
• Allow removal or replacement of objects within an image.
• Support replacement of an image's background.
• Enable removal of an image's background.

2. Support cost display

• Support for input/output token usage statistics and price display for each text model.
• Support for usage quantity statistics and price display for each image model.
• Support for calculating the total usage cost on the current device.

Note: Using New image-related features needs to upgrade your API, please refer: upgrade-api

Full Changelog: 1.6.0...1.7.0

What's Changed

Features

1. Support Amazon Nova models

• Stream conversations with Amazon Nova Micro, Lite and Pro
• Understand images, documents and videos with Nova Lite and Pro
• Record 30-second videos directly on Android and iOS for Nova analysis
• Upload large videos (1080p/4K) beyond 8MB with auto compression
• Support using Amazon Nova Canvas to generate images

Note: Video understanding needs to upgrade your API, please refer: upgrade-api

Full Changelog: 1.5.0...1.6.0