Introducing Zed AI, in collaboration with @anthropicAI.
Zed AI brings LLMs directly into your editor with an extensible, text-centric approach.
We're also piloting @anthropicAI's new Fast Edit mode for Claude 3.5 Sonnet with a small set of Zed users.
zed.dev/ai
Zed AI brings LLMs directly into your editor with an extensible, text-centric approach.
We're also piloting @anthropicAI's new Fast Edit mode for Claude 3.5 Sonnet with a small set of Zed users.
zed.dev/ai
Zed's assistant panel isn't a typical chat interface.
It's a full-fledged text editor exposing the entire LLM interaction.
Slash commands (/file, /terminal, /tab) integrate context, but every token is visible and editable.
Transparency and control, not magic.
It's a full-fledged text editor exposing the entire LLM interaction.
Slash commands (/file, /terminal, /tab) integrate context, but every token is visible and editable.
Transparency and control, not magic.
Zed's inline transformations let you edit code using natural language, with immediate, interactive feedback.
Activate with `ctrl-enter` and see changes instantly, thanks to our streaming diffs.
Powerful AI assistance while maintaining full control over the process.
Activate with `ctrl-enter` and see changes instantly, thanks to our streaming diffs.
Powerful AI assistance while maintaining full control over the process.
Fast Edit Mode: a breakthrough from @AnthropicAI that we're piloting in Zed.
It allows Claude 3.5 Sonnet to echo its input far faster than generating new text.
The result? Near-instantaneous refactoring.
We're collaborating with their team in Zed's open source codebase.
It allows Claude 3.5 Sonnet to echo its input far faster than generating new text.
The result? Near-instantaneous refactoring.
We're collaborating with their team in Zed's open source codebase.
Create custom slash commands via WebAssembly or our JSON-based protocol to integrate your own data sources, documentation, or proprietary information.
This lets you shape Zed's assistance to your exact needs.
This lets you shape Zed's assistance to your exact needs.
Use /workflow to direct multi-step transformations from the assistant panel.
It works well for Rust and Python, but it's still in alpha, so try it out and help shape its future!
It works well for Rust and Python, but it's still in alpha, so try it out and help shape its future!
Experience Zed AI for yourself!
It's as easy as logging into Zed with your GitHub account and opening the assistant panel.
It's as easy as logging into Zed with your GitHub account and opening the assistant panel.
For more information on Zed AI:
Read our in-depth blog post:
Explore the assistant documentation on our revamped website: zed.dev/blog/zed-ai
zed.dev/docs/assistant…
Read our in-depth blog post:
Explore the assistant documentation on our revamped website: zed.dev/blog/zed-ai
zed.dev/docs/assistant…
Today's stable (v0.149) release also includes:
- Added image support in the assistant panel (macOS only, currently).
- Added a count of the context tokens when hovering over token counts in the inline assist.
- Added support for GPT-4o in Copilot Chat.
- Added more context when using `assistant: Quote selection` to insert text into the assistant panel.
- Added support for Prompt Caching in Anthropic. For models that support it this can dramatically lower cost while improving performance.
- Added support for 8192 output tokens from Claude 3.5 Sonnet.
- Improved indentation when performing inline transformations.
- Zed now allows the model to control indentation when performing inline transformation. We're hoping this improves the indentation experience in Python and other indentation-sensitive languages, but it does require more from the model.
- Added `editor: scroll cursor {center,top,bottom}` actions for toggling scroll positions.
- Added an `editor::HandleInput` action to ease typing symbols that are part of keymaps.
- Vim: Added support for range syntax in commands.
- Vim: Prefill command with `:'<,'>` from visual mode.
- Added image support in the assistant panel (macOS only, currently).
- Added a count of the context tokens when hovering over token counts in the inline assist.
- Added support for GPT-4o in Copilot Chat.
- Added more context when using `assistant: Quote selection` to insert text into the assistant panel.
- Added support for Prompt Caching in Anthropic. For models that support it this can dramatically lower cost while improving performance.
- Added support for 8192 output tokens from Claude 3.5 Sonnet.
- Improved indentation when performing inline transformations.
- Zed now allows the model to control indentation when performing inline transformation. We're hoping this improves the indentation experience in Python and other indentation-sensitive languages, but it does require more from the model.
- Added `editor: scroll cursor {center,top,bottom}` actions for toggling scroll positions.
- Added an `editor::HandleInput` action to ease typing symbols that are part of keymaps.
- Vim: Added support for range syntax in commands.
- Vim: Prefill command with `:'<,'>` from visual mode.
Zed Preview v0.150 out too:
Today's preview release is largely identical to today's stable release. We've fast-tracked all new recently-released preview features to Zed stable, where they're currently being highlighted. There are a few subtle updates in preview, but the main action is happening in stable and around Zed AI. Next week's preview will be back on track with new features and improvements.
Today's preview release is largely identical to today's stable release. We've fast-tracked all new recently-released preview features to Zed stable, where they're currently being highlighted. There are a few subtle updates in preview, but the main action is happening in stable and around Zed AI. Next week's preview will be back on track with new features and improvements.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
