Q17 Private AI
Private offline AI chat, vision, voice and image generation
9 followers
Private offline AI chat, vision, voice and image generation
9 followers
Q17 Private AI brings private on-device AI to Android: offline AI chat, vision, voice input, offline TTS, file attachments, AI agents, separate project workspaces, benchmarks, CPU Safe Mode, and on-device image generation. It supports CPU/GPU backends with Vulkan and OpenCL acceleration on compatible phones, plus UI translations for 11 global languages.
Free Options
Launch Team / Built With
Tines The single, secure environment for agents, apps, and automations.
Promoted
how well does the on-device image generation actually hold up on mid-range phones without blowing up battery or throttling after a few prompts?
@nevzat1636963Fair question.
I built Q17 with real Android phones in mind, so I didn’t want local image generation to feel like a black box that just heats the device and leaves the user guessing.
It is still a heavy task, so I won’t pretend there is zero battery use or zero heat.
Q17 makes the process visible: progress %, denoise steps, elapsed time, ETA, and live thermal status are shown while the image is being generated.
The screenshot above shows the full flow: before generation → during generation → final output.
The device in this demo is a Poco M2 Pro, which is already an older non-flagship phone, not a new high-end demo device. Even there, Q17 is showing 63% progress, step 3/5, 49s elapsed, 43s ETA, and thermal status at 33.3°C marked Normal.
Q17 also has CPU Safe Mode and different presets, so weaker phones are not forced to behave like newer flagships. Older / low-RAM phones should stay on lighter modes, while stronger devices can push faster or more detailed runs.
Real-world speed depends on the phone, RAM, and preset. Very old 3GB RAM phones can take a few minutes, mid-range phones are more practical with lighter presets, and newer flagships can get much faster results.
So the goal is not “magic zero-cost generation.” The goal is private local image generation that stays transparent, stable, and usable on actual Android devices.
the CPU Safe Mode is a really thoughtful touch, especially for keeping things smooth on older devices while still letting people tap into local AI.
@sinansmekpcsd Thank you Sinan 🙌
CPU Safe Mode was important because not every phone behaves the same with local models. Some devices can use faster paths, but others need a safer fallback so the app stays stable instead of crashing or overheating.
That device-aware approach is a big part of Q17.