0.6.6

Changes

• hotfix: regression issue with colon in model name @louis-menlo (#6008)
• Add RunEvent::Exit event to tauri to handle macos context menu exit @qnixsynapse (#6005)
• fix: remove auto refresh model custom provider @urmauur (#6002)
• fix: generate response button disappear on tool call @louis-menlo (#5988)
• fix: title tooltip MCP edit json @urmauur (#5987)
• fix: download progress missing when left panel scrollable @urmauur (#5984)
• fix: failed provider models list due to broken cortex import @louis-menlo (#5983)
• Sync Release/v0.6.6 into dev @louis-menlo (#5973)
• fix: use direct process termination instead of console events on Windows @qnixsynapse (#5972)
• fix: rename thread dialog shows previous thread @urmauur (#5963)
• chore: allow all HTTPS image sources in img-src directive @Minh141120 (#5970)
• feat: Enhance port selection with availability check @qnixsynapse (#5966)
• fix: csp including img.shields.io and cdn-uploads.huggingface.co in img-src directive @Minh141120 (#5967)
• ci: tolerate artifact upload @Minh141120 (#5969)
• fix: assistant with last used and fix metadata @urmauur (#5955)
• fix: search models result in hub should be sorted by weight @louis-menlo (#5954)
• fix: factory reset fail with access denied error @louis-menlo (#5952)
• fix: set autoUnload in onLoad() @qnixsynapse (#5956)
• fix: update edge case experimental feature MCP @urmauur (#5951)
• fix: correctly apply auto_unload setting from config @qnixsynapse (#5953)
• fix: Prevent race condition with auto-unload during rapid model loading @qnixsynapse (#5947)
• chore: uninstall when upgrading windows installer @Minh141120 (#5945)
• fix: openrouter unselect itself @louis-menlo (#5943)
• fix: tool approval params scrollable @urmauur (#5941)
• fix: migrate app settings to the new version @louis-menlo (#5936)
• fix: Remove sInfo from activeSessions before unloading @qnixsynapse (#5938)
• fix: update default GPU toggle, and simplify state @urmauur (#5937)
• chore: revert back to passive mode on windows installer @Minh141120 (#5934)
• fix: update ui version_backend, mem usage hardware @urmauur (#5932)
• fix: Frontend updates when llama.cpp backend auto-downloads @qnixsynapse (#5926)
• fix: calculation memory on hardware and system monitor @urmauur (#5922)
• fix: persist model capabilities refresh app @urmauur (#5918)
• fix: validate name assistant and improve area clickable @urmauur (#5920)
• fix: Allow N-GPU Layers (NGL) to be set to 0 in llama.cpp @qnixsynapse (#5907)
• fix: models hub should show latest data only @louis-menlo (#5925)
• fix: Persist 'Auto-Unload Old Models' setting in llama.cpp @qnixsynapse (#5906)
• feat: Enhance Llama.cpp backend management with persistence @qnixsynapse (#5886)
• Chore cua mac runner @hiento09 (#5888)
• fix: provider settings should be refreshed on page load @louis-menlo (#5887)
• 🐛fix: get system info and system usage @urmauur (#5884)
• fix: gpu detected from backend version @urmauur (#5882)
• fix: bring back HF repo ID search in Hub @louis-menlo (#5880)
• chore: revert app artifact name for macos linux and windows builds @Minh141120 (#5878)
• feat: add support for querying available backend devices @qnixsynapse (#5877)
• fix: llama.cpp backend shows blank list sometime @louis-menlo (#5876)
• ci: rename app github artifact on windows and linux build @Minh141120 (#5875)
• ci: autoqa github artifact @Minh141120 (#5873)
• fix: jan should have a general assistant instruction @louis-menlo (#5872)
• fix: tmp download file should be removed on cancel @louis-menlo (#5849)
• 🐛fix: remove sampling parameters from llamacpp extension @urmauur (#5871)
• 🐛fix: update vulkan active syntax @urmauur (#5869)
• fix: app should not show manually deleted models @louis-menlo (#5868)
• feat: migrate cortex models to llamacpp extension @louis-menlo (#5838)
• fix: charmap encoding @Minh141120 (#5865)
• fix: HuggingFace provider should be non-deletable @louis-menlo (#5856)
• fix: gemini tool call support - version bump @louis-menlo (#5848)
• Fix: engine unable to find dlls on when running on Windows @qnixsynapse (#5863)
• chore: update build appimage script @Minh141120 (#5866)
• ✨enhancement: dialog model error trigger from provider screen @urmauur (#5858)
• fix: support load model configurations @urmauur (#5843)
• fix: delete all thread should not include fav @urmauur (#5864)
• Chore: enrich autoqa log @hiento09 (#5862)
• refactor: Improve Llama.cpp backend management and auto-update @qnixsynapse (#5845)
• fix: autoqa prompt template @Minh141120 (#5854)
• feat: add vcruntime for windows installer @Minh141120 (#5852)
• ✨enhancement: auto focus always allow action from tool approval dialog and add req parameters @urmauur (#5836)
• enhancement: better error page component @urmauur (#5834)
• chore: sync make build with dev @Minh141120 (#5847)
• refactor: standardize build process and remove build-tauri target @Minh141120 (#5846)
• fix: custom tauri nsis template CheckIfAppIsRunning macro @Minh141120 (#5840)
• fix: update @taur-apps/cli to newest verison to fix appimage download @Minh141120 (#5839)
• fix: prevent terminal window from opening on model load on WindowsOS @qnixsynapse (#5837)
• feat: add claude-4 @louis-menlo (#5829)
• feat: support per-model overrides in llama.cpp load() @qnixsynapse (#5820)
• fix: llama.cpp integration model load and chat experience @louis-menlo (#5823)
• test: deprecate webdriver test in favor of auto qa using CUA @louis-menlo (#5825)
• Revert "chore(deps): update rand requirement from 0.8 to 0.9 in /src-tauri" @louis-menlo (#5824)
• fix: Legacy threads show on top of new threads (#5696) @louis-menlo (#5810)
• fix: llama.cpp backend download on windows @louis-menlo (#5813)
• fix: dependabot should just update security patch @louis-menlo (#5814)
• chore(deps): update rand requirement from 0.8 to 0.9 in /src-tauri @dependabot (#5399)
• docs: Add Instruction for Toggling Experimental Features Before Toggling MCP Servers @bytrangle (#5771)
• chore(deps): bump @radix-ui/react-hover-card from 1.1.11 to 1.1.14 @dependabot (#5603)
• Fix autoqa lib dependencies @hiento09 (#5812)
• refactor: simplify proxy settings by removing unused SSL verification options @louis-menlo (#5809)
• feat: Add Hugging Face as a provider @gary149 (#5808)
• fix: Improve stream error handling and parsing @qnixsynapse (#5807)
• feat: add autoqa @hiento09 (#5779)
• set line number userSelect to none so that code can be copied without line number @ethanova (#5782)
• feat: add model load error handling to improve UX @louis-menlo (#5802)
• fix: Add --reasoning-format none to support rendering of reasoning content @qnixsynapse (#5803)
• feat: proxy support for the new downloader @louis-menlo (#5795)
• Sync release/0.6.5 into dev to start new development cycle @louis-menlo (#5801)
• refactor: move thinking toggle to runtime settings for dynamic control @qnixsynapse (#5800)
• test: deprecate webdriver test in favor of auto qa using CUA @louis-menlo (#5797)
• Documentation Updates for v0.6.5 @ramonpzg (#5799)

Contributor

@Minh141120, @bytrangle, @dependabot, @dependabot[bot], @ethanova, @gary149, @hiento09, @louis-menlo, @qnixsynapse, @ramonpzg and @urmauur