aicanrun
> 138 phones · 84 models · 514 quants

Can your phone run AI?

Yes — if you pick the right model. See what runs locally on your exact phone, how fast, and which app to use.

Check my phone →
iPhone 17 Pro MaxGalaxy S26 UltraPixel 10 ProNEWPixel 11 Pro XLNEWGalaxy Z Fold8 Ultra
transparent RAM math · transparent speed estimates · nothing to install
TRENDINGHow can I run DeepSeek Harness locally? Tested setup →DEEPSEEK V4 PROCan 849.7GB of weights run locally? →QWENQwen3.8 27B is open — can your 24GB or 32GB machine run it? →LING 3.0 FLASH124B total, 5.1B active — can your 64GB or 128GB machine run it? →
● LIVE CHECKS✓ Bonsai 27B (1-bit) on Galaxy Z Fold8✓ Bonsai 27B (1-bit) on iPhone 17 Pro Max✕ Qwen 3.6 35B-A3B on Galaxy S26 Ultra! Gemma 4 E4B on Pixel 10✕ Llama 3.3 70B on iPhone 17! Ternary Bonsai 8B on Nothing Phone (3a)! Qwen 3.5 9B on OnePlus 15! Ministral 3 14B on vivo X300 Pro✓ Qwen 3.5 4B on Galaxy S25 FE
01
Detect
Pick your phone in one tap — we already know the chip, RAM and memory bandwidth of 138 phones.
02
Match
Every model is checked with transparent RAM math — weights, KV cache and runtime overhead, per quant.
03
Run
Get the recommended quant, app and settings for your hardware — download and go.

Top models on phones

on Galaxy S26 Ultra ▾
GradeModelParamsNeedsTypical speed
SQwen3 0.6BQ8_0 · 0.6 GB0.6B1.3 GB~63.6 tokens/s›SQwen3 1.7BQ8_0 · 1.8 GB1.7B2.6 GB~21.2 tokens/s›SQwen3 4BQ4_K_M · 2.5 GB4B3.5 GB~15.3 tokens/s›SLlama 3.2 1BQ4_K_M · 0.8 GB1.2B1.5 GB~47.7 tokens/s›SLlama 3.2 3BQ4_K_M · 2 GB3.2B2.9 GB~19.1 tokens/s›SGemma 3 1BQ4_K_M · 0.8 GB1B1.5 GB~47.7 tokens/s›SGemma 3 4BQ4_K_M · 2.5 GB4.3B3.5 GB~15.3 tokens/s›SPhi-4 Mini 3.8BQ4_K_M · 2.5 GB3.8B3.5 GB~15.3 tokens/s›SDeepSeek R1 Distill 1.5BQ4_K_M · 1.1 GB1.8B1.9 GB~34.7 tokens/s›SSmolLM2 1.7BQ4_K_M · 1.1 GB1.7B1.9 GB~34.7 tokens/s›
see all 84 models →