kube vs Neuro

kube-coder mobile is the companion app for kube-coder, an open-source, self-hosted cloud development platform (github.com/imran31415/kube-coder). If you run a kube-coder workspace, this app lets you drive it from anywhere. CONNECT IN SECONDS Point the app at your workspace host and paste the API token from your dashboard (Settings → Mobile app). No account, no sign-up — the app talks directly to your own server, and the token is stored securely on your device. Just looking? Tap "Explore the public demo" on the first screen to browse a read-only demo workspace — no token needed. WHAT YOU CAN DO • Tasks — start a coding task for your AI assistant, follow the live session, and send follow-ups when it needs input. Active work up front; finished tasks tucked under Done. • Live terminal — the task view renders your workspace's real terminal, the same session the web dashboard shows. • Apps — see the dev servers running in your workspace and open them in-app. Split the screen to watch your app change while the assistant works on it. • Desktop — your one-tap launchers (build prompts, URLs, shell commands), shared with the web dashboard. • Memory — browse what your workspace remembers. • Metrics — live CPU, memory, and disk, plus service health. PRIVATE BY DESIGN No accounts. No analytics. No third-party services. The app connects only to the workspace host you configure, over HTTPS, using your token. kube-coder itself is open source — read the code, run your own, or try the demo: github.com/imran31415/kube-coder

Neuro lets you fine-tune real AI language models directly on your iPhone — no cloud, no servers, no data leaving your device. Pick a base model from the built-in catalog, choose a dataset (or bring your own), and train a specialized version tuned to your task. Everything runs locally using Apple's Neural Engine and GPU, powered by RWKV — an efficient architecture built for on-device use. FEATURES • Browse and download open models and datasets • Five training intensity levels, from light head-tuning to deep fine-tuning • Real-time training progress: loss, accuracy, memory usage • Fully offline — your data never leaves your device • Free and open source Neuro is built for people who want to experiment with real machine learning on real hardware, without renting a GPU or sending data anywhere. Whether you're a student, a researcher, or just curious how fine-tuning works, Neuro gives you a hands-on, private way to do it. This is an early release — expect rough edges, and expect it to get better. Feedback and contributions are welcome on GitHub.