Leaderboard
On-device LLM performance rankings powered by Glicko-2
Xiaomi 14T Pro
AndroidRank
#82
Rating
1,703
±17 RD
Win Rate
69.6%
Conservative Rating
1,669
TG Rating
1,735
PP Rating
1,505
Matches
921
Record
641W – 280L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| zephyr-smol_llama-100m-sft-full-Q8_0 | 164.14 | 1135.42 | 164.14 | 1135.42 | 1 |
| gemma-3-270m-it-F16 | 24.23 | 88.05 | 24.23 | 88.05 | 1 |
| llama-3.2-1b-instruct-q8_0 | 22.82 | 63.85 | 22.82 | 63.85 | 1 |
| SmolLM2-1.7B-Instruct-Q8_0 | 14.47 | 29.81 | 18.56 | 47.00 | 3 |
| lexitronLlama3.2-3B-Instruct.Q4_K_M | 12.68 | 19.28 | 12.68 | 19.28 | 1 |
| gemma-2-2b-it-Q6_K | 10.43 | 18.04 | 10.78 | 23.31 | 3 |
| Phi-3.5-mini-instruct.Q4_K_M | 9.27 | 18.47 | 10.59 | 23.21 | 2 |
| gemma-3-4b-it-Q8_0 | 7.67 | 21.01 | 7.67 | 21.01 | 1 |
| Gemmasutra-Mini-2B-v1-Q6_K | 6.85 | 19.22 | 6.85 | 19.22 | 1 |
| qwen2.5-3b-instruct-q5_k_m | 6.73 | 9.33 | 6.73 | 9.33 | 1 |
| gemma-3-1b-it.fp16 | 6.50 | 13.37 | 6.50 | 13.37 | 1 |
| Phi-4-mini-instruct-Q6_K | 4.48 | 11.81 | 4.48 | 11.81 | 1 |
| DeepSeek-R1-Distill-Qwen-7B-Q3_K_L | 3.37 | 4.85 | 3.37 | 4.85 | 1 |
Head-to-Head Record
1–50 of 265 rows
1 / 6
Performance by App Version
ImprovedRegressed