AI Models for ROG Phone 9 Pro — What runs on 24GB
Specs checked against manufacturer and public documentation on .
What runs on the ROG Phone 9 Pro
All 48 models at their recommended quant, on the 24GB configuration. Tap a model for the full report.
| Model | Params | Quant | Needs | Speed | Verdict | |
|---|---|---|---|---|---|---|
| S | Qwen3 0.6B | 0.6B | Q8_0 | 1.3 GB | ~57.6 tokens/s | ✓ Runs great |
| S | Qwen3 1.7B | 1.7B | Q8_0 | 2.6 GB | ~19.2 tokens/s | ✓ Runs great |
| S | Llama 3.2 1B | 1.2B | Q4_K_M | 1.5 GB | ~43.2 tokens/s | ✓ Runs great |
| S | Llama 3.2 3B | 3.2B | Q4_K_M | 2.9 GB | ~17.3 tokens/s | ✓ Runs great |
| S | Gemma 3 1B | 1B | Q4_K_M | 1.5 GB | ~43.2 tokens/s | ✓ Runs great |
| S | DeepSeek R1 Distill 1.5B | 1.8B | Q4_K_M | 1.9 GB | ~31.4 tokens/s | ✓ Runs great |
| S | SmolLM2 1.7B | 1.7B | Q4_K_M | 1.9 GB | ~31.4 tokens/s | ✓ Runs great |
| S | SmolLM3 3B | 3.1B | Q4_K_M | 2.8 GB | ~18.2 tokens/s | ✓ Runs great |
| S | Qwen 3.5 2B | 2B | Q4_K_M | 2.1 GB | ~26.6 tokens/s | ✓ Runs great |
| S | Gemma 4 26B-A4B | 26B | Q4_0 | 17.5 GB | ~15.6 tokens/s | ✓ Runs great |
| S | Ternary Bonsai 8B | 8B | PQ2_0 | 3.5 GB | ~15.7 tokens/s | ✓ Runs great |
| S | Ternary Bonsai 4B | 4B | PQ2_0 | 2 GB | ~31.4 tokens/s | ✓ Runs great |
| S | Ternary Bonsai 1.7B | 1.7B | PQ2_0 | 1.2 GB | ~69.1 tokens/s | ✓ Runs great |
| S | Ministral 3 3B | 3B | Q4_K_M | 3 GB | ~16.5 tokens/s | ✓ Runs great |
| S | LFM2.5 8B-A1B | 8B | Q4_K_M | 6.6 GB | ~53.2 tokens/s | ✓ Runs great |
| S | GPT-OSS 20B | 21B | MXFP4 | 14.8 GB | ~16.7 tokens/s | ✓ Runs great |
| S | Qwen3 4B | 4B | Q4_K_M | 3.5 GB | ~13.8 tokens/s | ✓ Runs great |
| S | Gemma 3 4B | 4.3B | Q4_K_M | 3.5 GB | ~13.8 tokens/s | ✓ Runs great |
| S | Phi-4 Mini 3.8B | 3.8B | Q4_K_M | 3.5 GB | ~13.8 tokens/s | ✓ Runs great |
| S | Qwen 3.5 4B | 4B | Q4_K_M | 3.7 GB | ~12.8 tokens/s | ✓ Runs great |
| S | Nemotron 3 Nano 4B | 4B | Q4_K_M | 3.8 GB | ~12.3 tokens/s | ✓ Runs great |
| S | Gemma 4 E2B | 2B | Q4_K_M | 4 GB | ~11.1 tokens/s | ✓ Runs great |
| A | Bonsai 27B (1-bit) | 27B | Q1_0 | 6.5 GB | ~9.1 tokens/s | ✓ Runs great |
| A | Mistral 7B v0.3 | 7.2B | Q4_K_M | 5.7 GB | ~7.9 tokens/s | ! Runs, barely |
| A | DeepSeek R1 Distill 7B | 7.6B | Q4_K_M | 6.1 GB | ~7.4 tokens/s | ! Runs, barely |
| A | Llama 3.1 8B | 8B | Q4_K_M | 6.3 GB | ~7.1 tokens/s | ! Runs, barely |
| A | Ministral 8B | 8B | Q4_K_M | 6.3 GB | ~7.1 tokens/s | ! Runs, barely |
| A | Qwen3 8B | 8.2B | Q4_K_M | 6.4 GB | ~6.9 tokens/s | ! Runs, barely |
| A | Gemma 4 E4B | 4B | Q4_K_M | 6.1 GB | ~6.9 tokens/s | ! Runs, barely |
| A | Ministral 3 8B | 8B | Q4_K_M | 6.6 GB | ~6.6 tokens/s | ! Runs, barely |
| A | Qwen 3.5 9B | 9B | Q4_K_M | 7.2 GB | ~6.1 tokens/s | ! Runs, barely |
| A | Gemma 3 12B | 12.2B | Q4_K_M | 9.1 GB | ~4.7 tokens/s | ! Runs, barely |
| A | Gemma 4 12B | 12B | Q4_0 | 8.8 GB | ~4.9 tokens/s | ! Runs, barely |
| A | Ternary Bonsai 27B | 27B | PQ2_0 | 10.1 GB | ~4.8 tokens/s | ! Runs, barely |
| A | Ministral 3 14B | 14B | Q4_K_M | 10.2 GB | ~4.2 tokens/s | ! Runs, barely |
| A | Qwen3 14B | 14.8B | Q4_K_M | 11.1 GB | ~3.8 tokens/s | ! Runs, barely |
| A | Phi-4 14B | 14.7B | Q4_K_M | 11.2 GB | ~3.8 tokens/s | ! Runs, barely |
| F | Qwen 3.5 27B | 27B | Q4_K_M | 20 GB | — | ✕ Won't fit |
| F | Qwen 3.6 27B | 27B | Q4_K_M | 20.1 GB | — | ✕ Won't fit |
| F | Qwen3 30B A3B | 30.5B | Q4_K_M | 22.3 GB | — | ✕ Won't fit |
| F | Gemma 4 31B | 31B | Q4_K_M | 22 GB | — | ✕ Won't fit |
| F | Qwen3 32B | 32.8B | Q4_K_M | 23.7 GB | — | ✕ Won't fit |
| F | Qwen 3.5 35B-A3B | 35B | Q4_K_M | 26.2 GB | — | ✕ Won't fit |
| F | Qwen 3.6 35B-A3B | 35B | Q4_K_M | 26.3 GB | — | ✕ Won't fit |
| F | Nemotron 3 Nano 30B-A3B | 30B | Q4_K_M | 28.5 GB | — | ✕ Won't fit |
| F | Llama 3.3 70B | 70B | Q4_K_M | 50.1 GB | — | ✕ Won't fit |
| F | Hunyuan 3 (Hy3) | 298.8B | Q4_K_M | 212.8 GB | — | ✕ Won't fit |
| F | Inkling | 952.4B | Q8_0 | 966.9 GB | — | ✕ Won't fit |
~ = bandwidth-based estimate · ✓ = measured on real hardware
Best model by use case
Top everyday assistant & writing pick here — ~57.6 tokens/s at Q8_0, using 1.3 of ~20GB.
Top code completion & explain-this pick here — ~13.8 tokens/s at Q4_K_M, using 3.5 of ~20GB.
Top math & step-by-step thinking pick here — ~31.4 tokens/s at Q4_K_M, using 1.9 of ~20GB.
FAQ
What is the biggest AI model the ROG Phone 9 Pro can run?
Bonsai 27B (1-bit) (27B parameters) at Q1_0 — it needs 6.5GB of the ~20GB usable on the 24GB ROG Phone 9 Pro and runs at ~9.1 tokens/s.
How much of the ROG Phone 9 Pro's 24GB RAM can AI models actually use?
About 20GB. Android keeps roughly 2–4GB for the system and resident apps, so of the 24GB about 20GB is actually available to a model.
Can the ROG Phone 9 Pro run Llama 3.1 8B?
Yes — at Q4_K_M it needs 6.3GB of the ~20GB usable and runs at ~7.1 tokens/s.
How fast is local AI on the ROG Phone 9 Pro?
The Snapdragon 8 Elite has 76.8GB/s of memory bandwidth, which is what decode speed scales with. Small models like Ternary Bonsai 1.7B reach ~69.1 tokens/s; larger 7–14B models land in the single digits. Anything above ~8 tokens/s feels smooth for chat.
Which quantization should I use on the ROG Phone 9 Pro?
Q4_K_M is the size/quality sweet spot for most models. For example, Qwen3 0.6B at Q8_0 takes 1.3GB of memory here. Only drop to Q3 or IQ4 if a model just misses fitting; Q8 rarely pays off on 24GB of RAM.
Is 24GB of RAM enough for local AI?
37 of the 48 models we track fit on the ROG Phone 9 Pro — 23 run great and 14 run with compromises. 11 models (mostly 12B+) don't fit at their recommended quant.