Leaderboard
On-device LLM performance rankings powered by Glicko-2
Galaxy Z Fold6
AndroidRank
#69
Rating
1,731
±15 RD
Win Rate
72.3%
Conservative Rating
1,700
TG Rating
1,748
PP Rating
1,697
Matches
1,151
Record
832W – 319L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| tinygemma3-Q8_0 | 313.80 | 1572.89 | 313.80 | 1572.89 | 1 |
| llama-3.2-1b-instruct-q8_0 | 28.18 | 112.18 | 29.10 | 114.09 | 3 |
| Llama-3.2-3B-Instruct-Q4_K_M | 14.40 | 27.89 | 14.40 | 27.89 | 1 |
| Gemmasutra-Mini-2B-v1-Q6_K | 12.56 | 26.00 | 13.90 | 29.23 | 5 |
| Qwen3-4B.Q4_K_M | 12.07 | 21.21 | 12.07 | 21.21 | 1 |
| gemma-2-2b-it-Q6_K | 11.00 | 27.25 | 14.41 | 39.83 | 6 |
| Llama-3.2-3B-Instruct-Q8_0 | 9.31 | 24.78 | 9.31 | 24.78 | 1 |
| qwen2.5-3b-instruct-q5_k_m | 8.91 | 15.76 | 9.05 | 16.08 | 2 |
| gemma-3-4b-it.Q5_K_S | 8.41 | 13.83 | 8.41 | 13.83 | 1 |
| Llama-3.2-3B-Instruct-Q6_K | 8.06 | 15.00 | 9.91 | 17.96 | 4 |
| Dolphin3-Llama3.2-Smart.Q8_0 | 7.40 | 25.61 | 7.40 | 25.61 | 1 |
| gemma3-4b-it-abliterated.Q4_K_M | 7.28 | 23.29 | 7.28 | 23.29 | 1 |
| Phi-3.5-mini-instruct.Q4_K_M | 4.79 | 16.79 | 4.79 | 16.79 | 1 |
| pygmalion-7b-Q5_K_M | 4.52 | 5.79 | 4.52 | 5.79 | 1 |
| criminal_investigation_QA.Q6_K | 4.08 | 10.00 | 4.08 | 10.00 | 1 |
| Deepseek-R1-Distill-NSFW-RPv1.Q4_K_M | 3.59 | 6.27 | 3.59 | 6.27 | 1 |
| Qwen3-4B-Instruct-2507-Q4_K_M | 3.52 | 14.38 | 3.52 | 14.38 | 1 |
| gemma-3-4b-it.Q8_0 | 3.50 | 16.71 | 3.50 | 16.71 | 1 |
| Qwen3-8B.Q3_K_L | 3.41 | 4.49 | 3.41 | 4.49 | 1 |
| Qwen3.5-2B-IQ4_NL | 2.82 | 108.85 | 2.82 | 108.85 | 1 |
Head-to-Head Record
1–50 of 287 rows
1 / 6
Performance by App Version
ImprovedRegressed