privateSLM: Local AI Chat

iOS Universel / Productivité

privateSLM — Private AI That Works Offline. No Subscription.

Stop renting AI. privateSLM runs powerful open-source AI models entirely on your iPhone, iPad and Mac — no account, no cloud, no monthly bill. Pay once, use it forever, and share it with your whole family.

WHY PRIVATESLM
• No subscription — one small payment, yours forever. No credits that “run out”. Cheaper than a single month of most AI apps.
• Works offline — on a plane, on the subway, travelling abroad with no data. Airplane mode is no problem.
• 100% private — everything runs on-device. No sign-up, no analytics, no ads, no tracking. Your chats are stored only on your phone.
• Family Sharing — one purchase covers up to six people across iPhone, iPad & Mac.

AN EXPERT FOR EVERY JOB
Download a small specialist model and run it locally:
• Coding & SQL — Qwen2.5 Coder, IBM Granite
• Math & STEM — Qwen2.5 Math, NuminaMath
• Medical — BioMistral, Meditron (research-tuned)
• Legal — SaulLM, Law Chat
• Finance — finance-tuned assistants
• Translation — TowerInstruct
• Mental-health & soft skills — MentaLLaMA
• Plus fast general chat models for everyday questions

BUILT TO ACTUALLY USE
• Matched to your iPhone — privateSLM checks your device's memory on-device and recommends models that will run well.
• Full chat history — multiple conversations you can rename and delete.
• Voice in and voice out — speak your question, hear the answer hands-free.
• Edit a message and regenerate. Stop, copy, or re-run any reply.
• Tune the system prompt and sampling (temperature, top-p, and more).
• Built-in model manager with resumable downloads.

RUNS ON APPLE INTELLIGENCE, TOO
On supported devices, privateSLM can use Apple's built-in system model — nothing to download, running on the Neural Engine.

PERFECT FOR
• Flights and travel with no connection
• When your other AI's credits have run out
• Anyone who wants private answers that never leave the phone
• Families who don't want five separate AI subscriptions

Runs on-device with llama.cpp. Your conversations stay on your phone — always.

Note: the Medical, Legal and Finance models are AI assistants for general information only — not professional medical, legal or financial advice. Larger 7B expert models run best on recent iPhones with 8 GB+ of memory.

Quoi de neuf dans la dernière version ?

This release is a full native rewrite — faster, lighter, and more private.
• Expert models by category: coding, math, medical, legal, finance, translation, mental-health, plus fast general chat.
• Model catalog matched to your iPhone's memory, checked on-device.
• Big 7B expert models now fit in memory reliably.
• Full chat history, voice input and spoken replies.
• Edit, regenerate, stop and copy replies; adjustable prompt & sampling.
• Uses Apple Intelligence on supported devices.