v0.7.5

• Support custom Azure OpenAI endpoint
• Make chat stick to bottom while generating response, unless you scroll up
• Fix: markdown headings line height
• Fix: incorrectly rendering :some_text in markdown

See the assets to download this version and install, or go to https://chatwise.app to download.

v0.7.4

• Set default max tokens of Claude 3.7 Sonnet to 8192 and the thinking mode to 64000
• Make Anthropic API compatible with AiHubMix (my aff link) (an alternative if you can't use the official Anthropic API)

See the assets to download this version and install, or go to https://chatwise.app to download.

v0.7.3

Add GitHub Copilit Claude 3.7 Sonnet (no thinking yet)

See the assets to download this version and install, or go to https://chatwise.app to download.

v0.7.2

Add support for Claude 3.7 Sonnet

See the assets to download this version and install, or go to https://chatwise.app to download.

v0.7.1

• Make @think work with GitHub Copilot Claude 3.5 Sonnet, Volcengine Doubao
• Less likely to trigger Google captcha when searching

See the assets to download this version and install, or go to https://chatwise.app to download.

v0.7.0

Web Search

• No more "Local browser" or "Fake browser", now you just choose a "search engine" and that's it, some search engines are implemented using API and others through web scrapping.
• Search engines: Google (free), Google (API), Jina (API), Exa (API), Brave Search (API), Tavily (API), learn more

Assistants

• Now we show the assistants in the left sidebar, the chats below is also filtered by selected assistant, the UX is still beta and misses some features, feedback is always welcome.

Other

• Markdown: add settings for markdown in Settings -> Advanced -> Markdown, currently you can configure inline LaTeX rendering
• @think: start your message with @think and it will first send messages to DeepSeek Reasoner to get the chain of thoughts, and then send the CoT to the current model to get answer. Experimental feature, might remove in the future. For example: try @think count r in strawberry in a non reasoning model

See the assets to download this version and install, or go to https://chatwise.app to download.

v0.6.91

Last release of this week, happy weekend!

Artifacts: allow to access microphone and camera, try this prompt: "make a html page that take camera input and render it as ascii"

See the assets to download this version and install, or go to https://chatwise.app to download.

v0.6.90

• Improve CMD+{number} shortcut
• Web search: rewrite the Fake browser provider in Rust, so it no longer needs to download Node.js to fetch Google results
• Web search: add a "Custom search" provider, you can provide your own search API: https://docs.chatwise.app/web-search.html#custom-search
• Support data export / import: only for settings, prompts and assistants currently, chat history is not yet supported

See the assets to download this version and install, or go to https://chatwise.app to download.

v0.6.89

Add quick switch for chat: hold Command (or Ctrl on Windows) for 0.5 second to switch to the last 10 chats by number

See the assets to download this version and install, or go to https://chatwise.app to download.

v0.6.88

• Set Azure OpenAI API version to 2024-12-01-preview, add an option in settings for this
• Make web search answer in user message's language, although sometimes models like gemini 2.0 flash will still answer in English
• Fix OpenRouter web search error
• Show changelog content when hovering the updater button

See the assets to download this version and install, or go to https://chatwise.app to download.