New Features & Capabilities
Internal Ollama Provider (v0.3.14)
- AMD GPU Support: Enabled automatically on macOS and during installation on Windows.
- Custom Model Imports: Import unlisted models via the default Ollama provider by pasting the import command.
- Ollama Library Example:
ollama run mistral-nemo
- Hugging Face Example:
ollama run hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF
Anthropic Computer Use (Experimental)
- Must be explicitly enabled in system settings. See documentation for details.
Workspace Chat
- Find-in-Page: Press
Ctrl+F to search text within the active workspace chat window.
Configuration & Integration Updates
LLM Providers & Inference
- NovitaAI: Added as a supported LLM provider.
- AWS Bedrock: Added support for Session Tokens.
API Updates
- Endpoint
workspace/chats: Added support for limit and orderBy parameters.
- File Uploads: Fixed API ingestion of files with non-latin characters.
Data & Document Ingestion
- Filetypes: Added support for
.ino files.
- Web Scraping: Resolved restriction blocking https://allm.loca.zone addresses.
- RAG: Improved document metadata for embedding results.
Bug Fixes
- Resolved ephemeral agent invocation bad reference.
- Resolved https://allm.loca.zone address block during web scraping.
- Resolved API upload failure for files containing non-latin characters.