Ternary Bonsai 4B
At the default Q2_0, 139 of 140 phones have enough estimated usable memory; 139 also reach 3+ tokens/s. On Mac, 54 of 54 exact configurations pass using that quant or a smaller tracked fallback.
Downloads & memory by quant
The download is only the weights — add KV cache and runtime to get what your phone actually needs. “Memory-fit” only means it can load; “usable” also requires an estimated 3+ tokens/s. Each row uses that row's quant. ★ = recommended.
| Quant | Download | Memory needed | Phone reality |
|---|---|---|---|
| Q2_0 ★ | 1.1 GB | ~2 GB | 139 memory-fit · 139 usable |
| Q2_0_g64 | 1.1 GB | ~2 GB | 139 memory-fit · 139 usable |
Ternary Bonsai 4B at Q2_0 on 140 phones
This table is fixed to the recommended Q2_0, on each phone's largest RAM option. Counts for smaller quants above will differ. All phones are shown, sorted by fit and estimated experience; tap one for the full report.
| Phone | Chip · RAM | Needs | Speed | Verdict |
|---|---|---|---|---|
| Galaxy S25 Ultra | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Galaxy S25 | Snapdragon 8 Elite · 12 GB | 2 / 9 GB | ~31.4 tokens/s | ✓ Runs great |
| Galaxy S24 Ultra | Snapdragon 8 Gen 3 · 12 GB | 2 / 9 GB | ~31.4 tokens/s | ✓ Runs great |
| Galaxy S24 | Samsung Exynos 2400 · 12 GB | 2 / 9 GB | ~27.5 tokens/s | ✓ Runs great |
| Galaxy S23 Ultra | Snapdragon 8 Gen 2 · 12 GB | 2 / 9 GB | ~27.5 tokens/s | ✓ Runs great |
| iPhone 16 Pro Max | Apple A18 Pro · 8 GB | 2 / 5.2 GB | ~24.5 tokens/s | ✓ Runs great |
| iPhone 16 Pro | Apple A18 Pro · 8 GB | 2 / 5.2 GB | ~24.5 tokens/s | ✓ Runs great |
| iPhone 16 | Apple A18 · 8 GB | 2 / 5.2 GB | ~24.5 tokens/s | ✓ Runs great |
| iPhone 15 Pro | Apple A17 Pro · 8 GB | 2 / 5.2 GB | ~20.9 tokens/s | ✓ Runs great |
| Pixel 9 Pro | Google Tensor G4 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 9 | Google Tensor G4 · 12 GB | 2 / 9 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 8 Pro | Google Tensor G3 · 12 GB | 2 / 9 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 8a | Google Tensor G3 · 8 GB | 2 / 6 GB | ~27.9 tokens/s | ✓ Runs great |
| OnePlus 13 | Snapdragon 8 Elite · 24 GB | 2 / 20 GB | ~31.4 tokens/s | ✓ Runs great |
| OnePlus 12 | Snapdragon 8 Gen 3 · 24 GB | 2 / 20 GB | ~31.4 tokens/s | ✓ Runs great |
| Xiaomi 15 Pro | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Xiaomi 14 | Snapdragon 8 Gen 3 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| POCO X6 Pro | Dimensity 8300 Ultra · 12 GB | 2 / 9 GB | ~27.9 tokens/s | ✓ Runs great |
| OPPO Find X8 Pro | Dimensity 9400 · 16 GB | 2 / 12 GB | ~34.9 tokens/s | ✓ Runs great |
| vivo X200 Pro | Dimensity 9400 · 16 GB | 2 / 12 GB | ~34.9 tokens/s | ✓ Runs great |
| vivo X100 Pro | Dimensity 9300 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| ROG Phone 9 Pro | Snapdragon 8 Elite · 24 GB | 2 / 20 GB | ~31.4 tokens/s | ✓ Runs great |
| Nothing Phone (3) | Snapdragon 8s Gen 4 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Motorola Edge 50 Ultra | Snapdragon 8s Gen 3 · 16 GB | 2 / 12 GB | ~27.5 tokens/s | ✓ Runs great |
| HONOR Magic7 Pro | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| iPhone 17 | Apple A19 · 8 GB | 2 / 5.2 GB | ~27.9 tokens/s | ✓ Runs great |
| iPhone Air | Apple A19 Pro · 12 GB | 2 / 7.8 GB | ~31.4 tokens/s | ✓ Runs great |
| iPhone 17 Pro | Apple A19 Pro · 12 GB | 2 / 7.8 GB | ~31.4 tokens/s | ✓ Runs great |
| iPhone 17 Pro Max | Apple A19 Pro · 12 GB | 2 / 7.8 GB | ~31.4 tokens/s | ✓ Runs great |
| Galaxy S26 | Snapdragon 8 Elite Gen 5 · 12 GB | 2 / 9 GB | ~34.7 tokens/s | ✓ Runs great |
| Galaxy S26+ | Snapdragon 8 Elite Gen 5 · 12 GB | 2 / 9 GB | ~34.7 tokens/s | ✓ Runs great |
| Galaxy S26 Ultra | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| Galaxy S26 FE | Exynos 2500 · 8 GB | 2 / 6 GB | ~31.4 tokens/s | ✓ Runs great |
| Xiaomi 17 | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| Xiaomi 17 Pro | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| OnePlus 15 | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| vivo X300 Pro | Dimensity 9500 · 16 GB | 2 / 12 GB | ~34.9 tokens/s | ✓ Runs great |
| vivo X300 E | Snapdragon 8 Gen 5 · 12 GB | 2 / 9 GB | ~31.4 tokens/s | ✓ Runs great |
| OPPO Find X9 Pro | Dimensity 9500 · 16 GB | 2 / 12 GB | ~34.9 tokens/s | ✓ Runs great |
| Pixel 10 | Google Tensor G5 · 12 GB | 2 / 9 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 10 Pro | Google Tensor G5 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 11 | Google Tensor G6 · 12 GB | 2 / 9 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 11 Pro | Google Tensor G6 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 11 Pro XL | Google Tensor G6 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| iPhone 16e | Apple A18 · 8 GB | 2 / 5.2 GB | ~24.5 tokens/s | ✓ Runs great |
| Pixel 9a | Google Tensor G4 · 8 GB | 2 / 6 GB | ~27.9 tokens/s | ✓ Runs great |
| Galaxy S25 FE | Samsung Exynos 2400 · 8 GB | 2 / 6 GB | ~27.5 tokens/s | ✓ Runs great |
| Galaxy Z Fold6 | Snapdragon 8 Gen 3 · 12 GB | 2 / 9 GB | ~31.4 tokens/s | ✓ Runs great |
| Galaxy Z Flip6 | Snapdragon 8 Gen 3 · 12 GB | 2 / 9 GB | ~31.4 tokens/s | ✓ Runs great |
| Galaxy Z Fold7 | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Galaxy Z Flip7 | Exynos 2500 · 12 GB | 2 / 9 GB | ~31.4 tokens/s | ✓ Runs great |
| Motorola Razr Ultra (2025) | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| OnePlus 13R | Snapdragon 8 Gen 3 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| OnePlus 15R | Snapdragon 8 Gen 5 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Pixel 9 Pro XL | Google Tensor G4 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 10 Pro XL | Google Tensor G5 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| iPhone 17e | Apple A19 · 8 GB | 2 / 5.2 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 10a | Google Tensor G4 · 8 GB | 2 / 6 GB | ~27.9 tokens/s | ✓ Runs great |
| Motorola Razr+ (2026) | Snapdragon 8s Gen 3 · 12 GB | 2 / 9 GB | ~27.5 tokens/s | ✓ Runs great |
| Motorola Razr Ultra (2026) | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| RedMagic 11 Air | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| RedMagic 11S Pro | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| Xiaomi 15T | Dimensity 8400 Ultra · 12 GB | 2 / 9 GB | ~27.9 tokens/s | ✓ Runs great |
| Xiaomi 15T Pro | Dimensity 9400+ · 12 GB | 2 / 9 GB | ~34.9 tokens/s | ✓ Runs great |
| POCO X7 Pro | Dimensity 8400 Ultra · 12 GB | 2 / 9 GB | ~27.9 tokens/s | ✓ Runs great |
| Xiaomi MIX Fold 2 | Snapdragon 8+ Gen 1 · 12 GB | 2 / 9 GB | ~20.9 tokens/s | ✓ Runs great |
| Honor Magic8 Pro | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| AQUOS R10 | Snapdragon 7+ Gen 3 · 12 GB | 2 / 9 GB | ~27.5 tokens/s | ✓ Runs great |
| Xperia 1 VII | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Xperia 1 VIII | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| Galaxy Z Fold8 | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| Galaxy Z Fold8 Ultra | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| Galaxy Z Flip8 | Samsung Exynos 2600 · 12 GB | 2 / 9 GB | ~34.9 tokens/s | ✓ Runs great |
| Galaxy Z Fold5 | Snapdragon 8 Gen 2 · 12 GB | 2 / 9 GB | ~27.5 tokens/s | ✓ Runs great |
| Galaxy Z Flip5 | Snapdragon 8 Gen 2 · 8 GB | 2 / 6 GB | ~27.5 tokens/s | ✓ Runs great |
| Xiaomi MIX Fold 3 | Snapdragon 8 Gen 2 · 16 GB | 2 / 12 GB | ~27.5 tokens/s | ✓ Runs great |
| Xiaomi MIX Fold 4 | Snapdragon 8 Gen 3 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Xiaomi MIX Flip | Snapdragon 8 Gen 3 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Xiaomi MIX Flip 2 | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Honor Magic V2 | Snapdragon 8 Gen 2 · 16 GB | 2 / 12 GB | ~27.5 tokens/s | ✓ Runs great |
| Honor Magic V3 | Snapdragon 8 Gen 3 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Honor Magic V5 | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| Honor Magic V6 | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| HONOR Robot Phone | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| OPPO Find N3 | Snapdragon 8 Gen 2 · 16 GB | 2 / 12 GB | ~27.5 tokens/s | ✓ Runs great |
| OnePlus Open | Snapdragon 8 Gen 2 · 16 GB | 2 / 12 GB | ~27.5 tokens/s | ✓ Runs great |
| OPPO Find N5 | Snapdragon 8 Elite · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| OPPO Find N6 | Snapdragon 8 Elite Gen 5 · 16 GB | 2 / 12 GB | ~34.7 tokens/s | ✓ Runs great |
| vivo X Fold3 | Snapdragon 8 Gen 2 · 16 GB | 2 / 12 GB | ~27.5 tokens/s | ✓ Runs great |
| vivo X Fold3 Pro | Snapdragon 8 Gen 3 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| vivo X Fold5 | Snapdragon 8 Gen 3 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| vivo X Fold6 | Dimensity 9500 · 16 GB | 2 / 12 GB | ~34.9 tokens/s | ✓ Runs great |
| Pixel Fold | Tensor G2 · 12 GB | 2 / 9 GB | ~20.9 tokens/s | ✓ Runs great |
| Pixel 9 Pro Fold | Google Tensor G4 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 10 Pro Fold | Google Tensor G5 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| Pixel 11 Pro Fold | Google Tensor G6 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| Huawei Mate X6 | Kirin 9020 · 16 GB | 2 / 12 GB | ~27.9 tokens/s | ✓ Runs great |
| Huawei Mate X7 | Kirin 9030 Pro · 20 GB | 2 / 16 GB | ~31.4 tokens/s | ✓ Runs great |
| Motorola Razr+ (2023) | Snapdragon 8+ Gen 1 · 12 GB | 2 / 9 GB | ~20.9 tokens/s | ✓ Runs great |
| Motorola Razr+ (2024) | Snapdragon 8s Gen 3 · 12 GB | 2 / 9 GB | ~27.5 tokens/s | ✓ Runs great |
| Motorola Razr+ (2025) | Snapdragon 8s Gen 3 · 12 GB | 2 / 9 GB | ~27.5 tokens/s | ✓ Runs great |
| Motorola Razr Fold | Snapdragon 8 Gen 5 · 16 GB | 2 / 12 GB | ~31.4 tokens/s | ✓ Runs great |
| iPhone 15 | Apple A16 Bionic · 6 GB | 2 / 3.9 GB | ~20.9 tokens/s | ✓ Runs great |
| Galaxy A57 5G | Exynos 1680 · 12 GB | 2 / 9 GB | ~14 tokens/s | ✓ Runs great |
| Nothing Phone (4a) Pro | Snapdragon 7 Gen 4 · 12 GB | 2 / 9 GB | ~13.7 tokens/s | ✓ Runs great |
| iPhone 14 | Apple A15 Bionic · 6 GB | 2 / 3.9 GB | ~14 tokens/s | ✓ Runs great |
| Galaxy A55 | Exynos 1480 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Redmi Note 13 Pro | Snapdragon 7s Gen 2 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Nothing Phone (2a) | Dimensity 7200 Pro · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Nothing Phone (3a) | Snapdragon 7s Gen 3 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Galaxy A26 5G | Exynos 1380 · 8 GB | 2 / 6 GB | ~10.5 tokens/s | ✓ Runs great |
| Galaxy A36 5G | Snapdragon 6 Gen 3 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Galaxy A56 5G | Exynos 1580 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Moto G Stylus (2025) | Snapdragon 6 Gen 3 · 8 GB | 2 / 6 GB | ~10.5 tokens/s | ✓ Runs great |
| Moto G Stylus (2026) | Snapdragon 6 Gen 3 · 8 GB | 2 / 6 GB | ~10.5 tokens/s | ✓ Runs great |
| Galaxy A37 5G | Exynos 1480 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Motorola Edge (2026) | Dimensity 7450 · 8 GB | 2 / 6 GB | ~10.5 tokens/s | ✓ Runs great |
| Redmi Note 14 Pro 5G | Dimensity 7300 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Redmi Note 15 Pro 5G | Dimensity 7400 Ultra · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Redmi Note 15 5G | Snapdragon 6 Gen 3 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Redmi Note 14 Pro+ 5G | Snapdragon 7s Gen 3 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Redmi Note 15 Pro+ 5G | Snapdragon 7s Gen 4 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| Honor 400 | Snapdragon 7 Gen 3 · 12 GB | 2 / 9 GB | ~10.5 tokens/s | ✓ Runs great |
| CMF Phone 2 Pro | Dimensity 7300 · 8 GB | 2 / 6 GB | ~10.5 tokens/s | ✓ Runs great |
| AQUOS sense10 | Snapdragon 7s Gen 3 · 8 GB | 2 / 6 GB | ~10.5 tokens/s | ✓ Runs great |
| AQUOS sense9 | Snapdragon 7s Gen 2 · 8 GB | 2 / 6 GB | ~10.5 tokens/s | ✓ Runs great |
| Xperia 10 VII | Snapdragon 6 Gen 3 · 8 GB | 2 / 6 GB | ~10.5 tokens/s | ✓ Runs great |
| Galaxy A16 5G | Exynos 1330 · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| Galaxy A17 5G | Exynos 1330 · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| Moto G (2025) | Dimensity 6300 · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| Moto G Power (2025) | Dimensity 6300 · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| Moto G Power (2026) | Dimensity 6300 · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| Motorola Razr (2025) | Dimensity 7400X · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| Motorola Razr (2026) | Dimensity 7450X · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| Motorola Razr (2024) | Dimensity 7300X · 12 GB | 2 / 9 GB | ~7 tokens/s | ! Runs, barely |
| Redmi Note 17 | Snapdragon 4 Gen 4 · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| vivo S2 | Dimensity 7360-Turbo · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| vivo V80 Lite 5G (Pakistan) | Dimensity 7360-Turbo · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| iQOO Z11 Lite | Dimensity 6300 · 8 GB | 2 / 6 GB | ~7 tokens/s | ! Runs, barely |
| Moto G (2026) | Dimensity 6300 · 4 GB | 2 / 2 GB | — | ✕ Won't fit |
~ = bandwidth-based estimate · ✓ = measured on real hardware
Ternary Bonsai 4B on 54 Mac configurations
Each Mac uses the recommended Q2_0 at 4K context when it fits, or the largest smaller tracked quant that fits. Each row links to the full report with a macOS-specific setup guide.
| Mac | Unified memory | Quant | Working budget | Needs | Estimated speed | Verdict |
|---|---|---|---|---|---|---|
| Mac Studio M5 Ultra | 512 GB | Q2_0 | 504 GB | 2.2 GB | ~545.5–763.6 tokens/s | ✓ Runs great |
| Mac Studio M3 Ultra | 512 GB | Q2_0 | 504 GB | 2.2 GB | ~372.3–521.2 tokens/s | ✓ Runs great |
| Mac Studio M5 Ultra | 256 GB | Q2_0 | 248 GB | 2.2 GB | ~545.5–763.6 tokens/s | ✓ Runs great |
| Mac Studio M3 Ultra | 256 GB | Q2_0 | 248 GB | 2.2 GB | ~372.3–521.2 tokens/s | ✓ Runs great |
| Mac Studio M5 Max | 128 GB | Q2_0 | 120 GB | 2.2 GB | ~279.1–390.7 tokens/s | ✓ Runs great |
| MacBook Pro M5 Max | 128 GB | Q2_0 | 120 GB | 2.2 GB | ~279.1–390.7 tokens/s | ✓ Runs great |
| Mac Studio M4 Max | 128 GB | Q2_0 | 120 GB | 2.2 GB | ~248.2–347.5 tokens/s | ✓ Runs great |
| MacBook Pro M4 Max | 128 GB | Q2_0 | 120 GB | 2.2 GB | ~248.2–347.5 tokens/s | ✓ Runs great |
| MacBook Pro M3 Max | 128 GB | Q2_0 | 120 GB | 2.2 GB | ~181.8–254.5 tokens/s | ✓ Runs great |
| Mac Studio M5 Ultra | 96 GB | Q2_0 | 88 GB | 2.2 GB | ~545.5–763.6 tokens/s | ✓ Runs great |
| Mac Studio M3 Ultra | 96 GB | Q2_0 | 88 GB | 2.2 GB | ~372.3–521.2 tokens/s | ✓ Runs great |
| Mac mini M5 Pro | 64 GB | Q2_0 | 57.6 GB | 2.2 GB | ~139.5–195.4 tokens/s | ✓ Runs great |
| Mac Studio M5 Max | 64 GB | Q2_0 | 57.6 GB | 2.2 GB | ~279.1–390.7 tokens/s | ✓ Runs great |
| MacBook Pro M5 Pro | 64 GB | Q2_0 | 57.6 GB | 2.2 GB | ~139.5–195.4 tokens/s | ✓ Runs great |
| MacBook Pro M5 Max | 64 GB | Q2_0 | 57.6 GB | 2.2 GB | ~279.1–390.7 tokens/s | ✓ Runs great |
| Mac Studio M4 Max | 64 GB | Q2_0 | 57.6 GB | 2.2 GB | ~248.2–347.5 tokens/s | ✓ Runs great |
| Mac mini M4 Pro | 64 GB | Q2_0 | 57.6 GB | 2.2 GB | ~124.1–173.7 tokens/s | ✓ Runs great |
| MacBook Pro M4 Max | 64 GB | Q2_0 | 57.6 GB | 2.2 GB | ~248.2–347.5 tokens/s | ✓ Runs great |
| MacBook Pro M3 Max | 64 GB | Q2_0 | 57.6 GB | 2.2 GB | ~181.8–254.5 tokens/s | ✓ Runs great |
| Mac mini M5 Pro | 48 GB | Q2_0 | 43.2 GB | 2.2 GB | ~139.5–195.4 tokens/s | ✓ Runs great |
| Mac Studio M5 Max | 48 GB | Q2_0 | 43.2 GB | 2.2 GB | ~279.1–390.7 tokens/s | ✓ Runs great |
| MacBook Pro M5 Pro | 48 GB | Q2_0 | 43.2 GB | 2.2 GB | ~139.5–195.4 tokens/s | ✓ Runs great |
| Mac Studio M4 Max | 48 GB | Q2_0 | 43.2 GB | 2.2 GB | ~248.2–347.5 tokens/s | ✓ Runs great |
| Mac mini M4 Pro | 48 GB | Q2_0 | 43.2 GB | 2.2 GB | ~124.1–173.7 tokens/s | ✓ Runs great |
| MacBook Pro M4 Pro | 48 GB | Q2_0 | 43.2 GB | 2.2 GB | ~124.1–173.7 tokens/s | ✓ Runs great |
| Mac Studio M5 Max | 36 GB | Q2_0 | 32.4 GB | 2.2 GB | ~209.1–292.7 tokens/s | ✓ Runs great |
| MacBook Pro M5 Max | 36 GB | Q2_0 | 32.4 GB | 2.2 GB | ~209.1–292.7 tokens/s | ✓ Runs great |
| Mac Studio M4 Max | 36 GB | Q2_0 | 32.4 GB | 2.2 GB | ~186.4–260.9 tokens/s | ✓ Runs great |
| MacBook Pro M4 Max | 36 GB | Q2_0 | 32.4 GB | 2.2 GB | ~186.4–260.9 tokens/s | ✓ Runs great |
| MacBook Pro M3 Pro | 36 GB | Q2_0 | 32.4 GB | 2.2 GB | ~68.2–95.5 tokens/s | ✓ Runs great |
| MacBook Pro M3 Max | 36 GB | Q2_0 | 32.4 GB | 2.2 GB | ~136.4–190.9 tokens/s | ✓ Runs great |
| Mac mini M6 | 32 GB | Q2_0 | 28.8 GB | 2.2 GB | ~77.3–108.2 tokens/s | ✓ Runs great |
| Mac mini M5 Pro | 32 GB | Q2_0 | 28.8 GB | 2.2 GB | ~139.5–195.4 tokens/s | ✓ Runs great |
| MacBook Air M5 | 32 GB | Q2_0 | 28.8 GB | 2.2 GB | ~58.4–80.7 tokens/s | ✓ Runs great |
| MacBook Air M4 | 32 GB | Q2_0 | 28.8 GB | 2.2 GB | ~45.8–63.3 tokens/s | ✓ Runs great |
| Mac mini M4 | 32 GB | Q2_0 | 28.8 GB | 2.2 GB | ~54.5–76.4 tokens/s | ✓ Runs great |
| Mac mini M6 | 24 GB | Q2_0 | 21 GB | 2.2 GB | ~77.3–108.2 tokens/s | ✓ Runs great |
| Mac mini M5 Pro | 24 GB | Q2_0 | 21 GB | 2.2 GB | ~139.5–195.4 tokens/s | ✓ Runs great |
| MacBook Air M5 | 24 GB | Q2_0 | 21 GB | 2.2 GB | ~58.4–80.7 tokens/s | ✓ Runs great |
| MacBook Pro M5 Pro | 24 GB | Q2_0 | 21 GB | 2.2 GB | ~139.5–195.4 tokens/s | ✓ Runs great |
| MacBook Air M4 | 24 GB | Q2_0 | 21 GB | 2.2 GB | ~45.8–63.3 tokens/s | ✓ Runs great |
| Mac mini M4 | 24 GB | Q2_0 | 21 GB | 2.2 GB | ~54.5–76.4 tokens/s | ✓ Runs great |
| Mac mini M4 Pro | 24 GB | Q2_0 | 21 GB | 2.2 GB | ~124.1–173.7 tokens/s | ✓ Runs great |
| MacBook Air M3 | 24 GB | Q2_0 | 21 GB | 2.2 GB | ~38.2–52.7 tokens/s | ✓ Runs great |
| MacBook Pro M4 Pro | 24 GB | Q2_0 | 21 GB | 2.2 GB | ~124.1–173.7 tokens/s | ✓ Runs great |
| MacBook Pro M3 Pro | 18 GB | Q2_0 | 15 GB | 2.2 GB | ~68.2–95.5 tokens/s | ✓ Runs great |
| Mac mini M6 | 16 GB | Q2_0 | 13 GB | 2.2 GB | ~77.3–108.2 tokens/s | ✓ Runs great |
| MacBook Air M5 | 16 GB | Q2_0 | 13 GB | 2.2 GB | ~58.4–80.7 tokens/s | ✓ Runs great |
| MacBook Air M4 | 16 GB | Q2_0 | 13 GB | 2.2 GB | ~45.8–63.3 tokens/s | ✓ Runs great |
| Mac mini M4 | 16 GB | Q2_0 | 13 GB | 2.2 GB | ~54.5–76.4 tokens/s | ✓ Runs great |
| MacBook Air M3 | 16 GB | Q2_0 | 13 GB | 2.2 GB | ~38.2–52.7 tokens/s | ✓ Runs great |
| MacBook Air M1 | 16 GB | Q2_0 | 13 GB | 2.2 GB | ~26.1–36 tokens/s | ✓ Runs great |
| MacBook Neo A18 Pro | 8 GB | Q2_0 | 5 GB | 2.2 GB | ~22.9–31.6 tokens/s | ✓ Runs great |
| MacBook Air M1 | 8 GB | Q2_0 | 5 GB | 2.2 GB | ~26.1–36 tokens/s | ✓ Runs great |
FAQ
How much RAM do I need to run Ternary Bonsai 4B on a phone?
At Q2_0, Ternary Bonsai 4B needs ~2GB of usable memory (weights + KV cache + runtime). In practice that means a 6GB+ Android phone or a 6GB+ iPhone — iOS lets a single app use less of its RAM than Android does.
How fast does Ternary Bonsai 4B run on a flagship phone?
On the Galaxy S25 Ultra (Snapdragon 8 Elite) it runs at ~31.4 tokens/s at Q2_0, estimated from memory bandwidth. Anything above ~8 tokens/s feels smooth for chat.
Can an iPhone run Ternary Bonsai 4B?
Yes — the iPhone 16 Pro Max runs it at ~24.5 tokens/s at Q2_0, using 2GB of its ~5.2GB usable memory.
What is the best quantization of Ternary Bonsai 4B for mobile?
Q2_0 (1.1GB download) is the size/quality sweet spot of the 2 quants available. Total memory needed is ~2GB once the KV cache and runtime are counted — the download size alone understates it.
How many phones can run Ternary Bonsai 4B?
139 of the 140 phones we track have enough estimated usable memory at Q2_0; 139 also reach our usable-speed threshold of 3 tokens/s, and 127 run smoothly. Memory-fit alone does not mean a good experience.
Which Macs can run Ternary Bonsai 4B?
54 of the 54 exact Mac configurations we track have enough estimated working memory. Each Mac uses the recommended Q2_0 when it fits, or the largest smaller tracked quant that fits, and separates installed unified memory, the macOS reserve, and estimated speed.
What is Ternary Bonsai 4B good for on a phone?
It's tagged for chat. At 4B parameters, it's a fast, lightweight pick for quick tasks.