Leaderboard
On-device LLM performance rankings powered by Glicko-2
POCO X3 Pro
AndroidRank
#187
Rating
1,331
±20 RD
Win Rate
33.6%
Conservative Rating
1,290
TG Rating
1,351
PP Rating
1,253
Matches
669
Record
225W – 444L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| deepseek-r1-distill-qwen-1.5b-uncensored-q4_k_s | 10.35 | 28.10 | 10.35 | 28.10 | 1 |
| Yi-Coder-1.5B-Chat.Q5_K_M | 9.93 | 23.25 | 9.93 | 23.25 | 1 |
| DeepSeek-R1-Distill-Qwen-1.5B-uncensored.i1-Q6_K | 7.77 | 17.91 | 7.77 | 17.91 | 1 |
| DeepSeek-R1-Distill-Qwen-1.5B-Q8_0 | 7.18 | 26.81 | 7.18 | 26.81 | 1 |
| DeepSeek-R1-Distill-Qwen-1.5B-IQ2_M | 7.07 | 10.89 | 7.07 | 10.89 | 1 |
| DeepSeek-R1-Distill-Qwen-1.5B-IQ3_XS | 5.91 | 9.23 | 5.91 | 9.23 | 1 |
| gemma-3n-E2B-it-Q4_K_M | 5.58 | 17.33 | 5.65 | 17.58 | 2 |
| PowerInfer.SmallThinker-3B-Preview.Q4_K_M | 5.46 | 13.32 | 5.46 | 13.32 | 1 |
| Phi-3.5-mini-instruct.Q4_K_M | 5.13 | 8.71 | 5.13 | 8.71 | 1 |
| Llama-3.2-3B-Agent007-Coder.Q5_K_M | 5.03 | 9.57 | 5.03 | 9.57 | 1 |
| neutss-air-BF16 | 4.66 | 13.99 | 4.66 | 13.99 | 1 |
| gemma-2-2b-it-Q6_K | 4.42 | 11.10 | 4.42 | 11.10 | 1 |
| Qwen3.5-0.8B-BF16 | 4.16 | 14.57 | 4.16 | 14.57 | 1 |
| Llama-3.2-3B-Instruct-Q6_K | 3.11 | 8.25 | 3.93 | 8.80 | 3 |
Head-to-Head Record
1–50 of 188 rows
1 / 4
Performance by App Version
ImprovedRegressed