Highlights
Faster Conversations
- Re-opening a conversation no longer pays for refetching every long tool output up front — full tool content is loaded on demand when you expand a truncated tool call, so big histories open noticeably faster.
- Removed redundant warmup and MCP sync work on the message-send path so the first reply in a conversation arrives sooner, especially on aionrs and ACP backends.
Model Selector Reliability
- Switching the model inside a conversation now sticks: leaving and returning to the conversation keeps your chosen model, and the home page dropdown remembers the last choice across ACP, aionrs, and team views.
Bug Fixes
- Packaged installs that ship without bundled AionCore now show a clear recovery screen with a download link and diagnostic report action instead of failing silently; release builds also verify the bundled runtime during packaging so this can't ship broken again.
- Provider health checks in Settings now use a dedicated probe and surface the backend's real error message — no more stale "error" text or temporary conversation rows polluting your sidebar.
- Cross-architecture builds (e.g. Mac x64 from an arm runner) now bundle the matching AionCore binary instead of the runner's architecture.
What's Changed
- fix(backend): handle incomplete bundled aioncore installs by @piorpua in #3078
- fix(settings): use health check error message by @piorpua in #3080
- Avoid redundant warmup and MCP sync during message startup by @piorpua in #3082
- fix(model-selector): trust backend current model and persist user preferences by @TCP404 in #3084
- Lazy-load full tool message content by @piorpua in #3086
- fix(settings): use provider health check probe by @piorpua in #3090
- fix(build): align bundled aioncore target arch by @piorpua in #3092
- chore: bump version to 2.1.6 by @piorpua in #3094
Full Changelog: v2.1.5...v2.1.6