NameQuick 2.11.26 – Release Notes

Release date: 2026-05-04

New

Added stronger control for keeping MLX models loaded. When you choose to keep a local MLX model loaded, NameQuick now treats that as an app-level promise while the app is open.

Made local model wording easier to understand. MLX controls now use familiar language like “Load model,” “Unload model,” and “Keep model loaded,” matching how Ollama and LM Studio users already think about local models.
Improved MLX setup and readiness checks. NameQuick now handles the bundled runtime environment more carefully, checks that the runtime is actually installed before starting, and keeps setup progress from being overwritten mid-install.
Made the MLX settings control more compact. The keep-loaded switch now takes up less space while staying easy to scan in the local AI settings.
Kept self-managed AI routing more consistent. BYOK processing now preserves the selected provider context more reliably during file preparation and rename operations.

Fixed MLX models unloading unexpectedly. Normal cleanup, idle release, and delayed release paths now respect the keep-loaded setting instead of stopping the model behind the scenes.
Fixed local MLX startup getting stuck after stale runtime state. NameQuick now refreshes runtime availability more carefully so installed models and runtime status stay in sync.
Fixed settings refresh loops around MLX configuration. Opening and updating the local AI settings should be calmer, with fewer repeated state updates while the app checks model and runtime readiness.