MLXBar v1.6.2 is available.
API-requested models can now stay resident together under per-model and aggregate memory limits.
INDEPENDENT DEVELOPER · HIROSHIMA
Make the worldmore usefulmore interesting
I build so that everyone has a chance to experience
what technology can make easier and more fun.
BUILDING NOWIdeas into products
0OPEN PROJECTS
01LATEST ARTICLES
Gain insight
↗
02PROFILE
Meet the person
behind the work.
→
NEWS & UPDATES
What is new
starts here
↓PROJECT ARCHIVE
Things I made.
Things I keep growing.
Music, learning, AI, and file management. I find small everyday frictions and turn them into products you can actually use.
01Awasero MusicTurn humming into editable scores and MIDI.macOSOPEN ↗02EasyRooAutomate daily work with a local LLM.macOSOPEN ↗03KiziArticles about software development and AI.WebOPEN ↗04MCS ManagerManage Minecraft servers from one place.macOSOPEN ↗05MD Viewer ProView and edit Markdown, then export PDF or HTML.macOSOPEN ↗06Media MasterManage photos, music, and documents together.AndroidOPEN ↗07MLXBarLocal AI for Mac, right from the menu bar.macOSOPEN ↗08Pine ChatAn autonomous local AI agent for macOS.macOSOPEN ↗09Tango proA vocabulary app built around your own CSV.AndroidmacOSOPEN ↗10Vocello JPA Japanese AI voice studio that stays on Mac.macOSOPEN ↗11Volume RoutineAutomate Android volume by time and Wi-Fi.AndroidOPEN ↗12WAKARUTurn documents into cited understanding.macOSOPEN ↗
NEWS / 2026
News
↗
MLXBar v1.6.1 is available.
Prompt reuse across branching conversations now actually engages on the models it was built for.
MLXBar v1.6.0 is available.
Cancelling a generation now keeps the work already done, so the next turn resumes instead of starting over.
HELLO, WORLD.