Angaben zum Datum
Datum aus der Quelle.
Erstmals gesehen am .
Raycast: Weniger Speicherverbrauch und neue Wege, Kontext an AI zu senden
AI Chat, Notes und Settings laufen nun jeweils in einem eigenen Prozess, der Speicherverbrauch von Launcher und File Search sinkt deutlich, die File Search wird schneller, und die neuen Befehle Send Element to AI und Send Window to AI senden Bildschirminhalte an AI Chat; zudem können Custom AI Providers Tools wie Web Search nutzen und Script Commands lassen sich einzeln für Hintergrundaktualisierung steuern.
image
We made more memory and performance improvements. AI Chat, Notes and Settings now each run in their own process, so a busy chat or a heavy note no longer slows down Root Search, and closing them frees up their memory. The launcher can use up to 35% less memory while idle. File Search uses up to 80% less memory at peak and up to 95% less while searching, searches are up to 30% faster, and background indexing uses much less CPU. More details are below.
💬 Send Context to AI
Two new commands let you show AI exactly what you're looking at. Send Element to AI lets you hover over anything on your screen, like a paragraph, an image or a button, and click to send it to AI Chat. To share a whole window instead, use Send Window to AI.
Tip: Assign a hotkey like ⌘ + ⇧ + 0 to Send Element to AI to grab context even faster.
✨ New
- Custom AI Providers: You can now give custom provider models Raycast's Web Search, Read Page, Search Images and Chart tools. Turn them on with a
toolsblock inproviders.yaml, and choose whether each one runs on Raycast, your company gateway, or a vendor you have a key for (Tavily, Exa, Firecrawl, Brave, Perplexity or Jina) - Script Commands: Added controls to enable or disable background refresh for individual scripts