AI Models for Xperia 10 VII — What runs on 8GB
Specs checked against manufacturer and public documentation on .
What runs on the Xperia 10 VII
All 48 models at their recommended quant, on the 8GB 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 | ~19.2 tokens/s | ✓ Runs great |
| S | Ternary Bonsai 1.7B | 1.7B | PQ2_0 | 1.2 GB | ~23 tokens/s | ✓ Runs great |
| S | Llama 3.2 1B | 1.2B | Q4_K_M | 1.5 GB | ~14.4 tokens/s | ✓ Runs great |
| S | Gemma 3 1B | 1B | Q4_K_M | 1.5 GB | ~14.4 tokens/s | ✓ Runs great |
| S | DeepSeek R1 Distill 1.5B | 1.8B | Q4_K_M | 1.9 GB | ~10.5 tokens/s | ✓ Runs great |
| S | SmolLM2 1.7B | 1.7B | Q4_K_M | 1.9 GB | ~10.5 tokens/s | ✓ Runs great |
| S | Ternary Bonsai 4B | 4B | PQ2_0 | 2 GB | ~10.5 tokens/s | ✓ Runs great |
| A | Qwen 3.5 2B | 2B | Q4_K_M | 2.1 GB | ~8.9 tokens/s | ✓ Runs great |
| A | Qwen3 1.7B | 1.7B | Q8_0 | 2.6 GB | ~6.4 tokens/s | ! Runs, barely |
| A | SmolLM3 3B | 3.1B | Q4_K_M | 2.8 GB | ~6.1 tokens/s | ! Runs, barely |
| A | Llama 3.2 3B | 3.2B | Q4_K_M | 2.9 GB | ~5.8 tokens/s | ! Runs, barely |
| A | Ministral 3 3B | 3B | Q4_K_M | 3 GB | ~5.5 tokens/s | ! Runs, barely |
| A | Ternary Bonsai 8B | 8B | PQ2_0 | 3.5 GB | ~5.2 tokens/s | ! Runs, barely |
| A | Qwen3 4B | 4B | Q4_K_M | 3.5 GB | ~4.6 tokens/s | ! Runs, barely |
| A | Gemma 3 4B | 4.3B | Q4_K_M | 3.5 GB | ~4.6 tokens/s | ! Runs, barely |
| A | Phi-4 Mini 3.8B | 3.8B | Q4_K_M | 3.5 GB | ~4.6 tokens/s | ! Runs, barely |
| A | Qwen 3.5 4B | 4B | Q4_K_M | 3.7 GB | ~4.3 tokens/s | ! Runs, barely |
| A | Nemotron 3 Nano 4B | 4B | Q4_K_M | 3.8 GB | ~4.1 tokens/s | ! Runs, barely |
| A | Gemma 4 E2B | 2B | Q4_K_M | 4 GB | ~3.7 tokens/s | ! Runs, barely |
| C | Mistral 7B v0.3 | 7.2B | Q4_K_M | 5.7 GB | ~2.6 tokens/s | ! Runs, barely |
| F | DeepSeek R1 Distill 7B | 7.6B | Q4_K_M | 6.1 GB | — | ✕ Won't fit |
| F | Gemma 4 E4B | 4B | Q4_K_M | 6.1 GB | — | ✕ Won't fit |
| F | Qwen3 8B | 8.2B | Q4_K_M | 6.4 GB | — | ✕ Won't fit |
| F | Llama 3.1 8B | 8B | Q4_K_M | 6.3 GB | — | ✕ Won't fit |
| F | Ministral 8B | 8B | Q4_K_M | 6.3 GB | — | ✕ Won't fit |
| F | Bonsai 27B (1-bit) | 27B | Q1_0 | 6.5 GB | — | ✕ Won't fit |
| F | Ministral 3 8B | 8B | Q4_K_M | 6.6 GB | — | ✕ Won't fit |
| F | LFM2.5 8B-A1B | 8B | Q4_K_M | 6.6 GB | — | ✕ Won't fit |
| F | Qwen 3.5 9B | 9B | Q4_K_M | 7.2 GB | — | ✕ Won't fit |
| F | Gemma 4 12B | 12B | Q4_0 | 8.8 GB | — | ✕ Won't fit |
| F | Gemma 3 12B | 12.2B | Q4_K_M | 9.1 GB | — | ✕ Won't fit |
| F | Ternary Bonsai 27B | 27B | PQ2_0 | 10.1 GB | — | ✕ Won't fit |
| F | Ministral 3 14B | 14B | Q4_K_M | 10.2 GB | — | ✕ Won't fit |
| F | Qwen3 14B | 14.8B | Q4_K_M | 11.1 GB | — | ✕ Won't fit |
| F | Phi-4 14B | 14.7B | Q4_K_M | 11.2 GB | — | ✕ Won't fit |
| F | Qwen3 30B A3B | 30.5B | Q4_K_M | 22.3 GB | — | ✕ Won't fit |
| F | Qwen 3.5 27B | 27B | Q4_K_M | 20 GB | — | ✕ Won't fit |
| F | Qwen 3.5 35B-A3B | 35B | Q4_K_M | 26.2 GB | — | ✕ Won't fit |
| F | Qwen 3.6 27B | 27B | Q4_K_M | 20.1 GB | — | ✕ Won't fit |
| F | Qwen 3.6 35B-A3B | 35B | Q4_K_M | 26.3 GB | — | ✕ Won't fit |
| F | Gemma 4 26B-A4B | 26B | Q4_0 | 17.5 GB | — | ✕ Won't fit |
| F | Nemotron 3 Nano 30B-A3B | 30B | Q4_K_M | 28.5 GB | — | ✕ Won't fit |
| F | GPT-OSS 20B | 21B | MXFP4 | 14.8 GB | — | ✕ Won't fit |
| F | Llama 3.3 70B | 70B | Q4_K_M | 50.1 GB | — | ✕ Won't fit |
| F | Qwen3 32B | 32.8B | Q4_K_M | 23.7 GB | — | ✕ Won't fit |
| F | Gemma 4 31B | 31B | Q4_K_M | 22 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 — ~19.2 tokens/s at Q8_0, using 1.3 of ~6GB.
Top code completion & explain-this pick here — ~4.6 tokens/s at Q4_K_M, using 3.5 of ~6GB.
Top math & step-by-step thinking pick here — ~10.5 tokens/s at Q4_K_M, using 1.9 of ~6GB.
FAQ
What is the biggest AI model the Xperia 10 VII can run?
Ternary Bonsai 8B (8B parameters) at PQ2_0 — it needs 3.5GB of the ~6GB usable on the 8GB Xperia 10 VII, at ~5.2 tokens/s.
How much of the Xperia 10 VII's 8GB RAM can AI models actually use?
About 6GB. Android keeps roughly 2–4GB for the system and resident apps, so of the 8GB about 6GB is actually available to a model.
Can the Xperia 10 VII run Llama 3.1 8B?
Not at Q4_K_M: it needs 6.3GB but the Xperia 10 VII only has ~6GB usable. Try a smaller model like Qwen3 0.6B.
How fast is local AI on the Xperia 10 VII?
The Snapdragon 6 Gen 3 has 25.6GB/s of memory bandwidth, which is what decode speed scales with. Small models like Ternary Bonsai 1.7B reach ~23 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 Xperia 10 VII?
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 8GB of RAM.
Is 8GB of RAM enough for local AI?
20 of the 48 models we track fit on the Xperia 10 VII — 8 run great and 12 run with compromises. 28 models (mostly 12B+) don't fit at their recommended quant.