Leaderboard
On-device LLM performance rankings powered by Glicko-2
X200 Pro
AndroidRank
#55
Rating
1,767
±14 RD
Win Rate
75.7%
Conservative Rating
1,738
TG Rating
1,786
PP Rating
1,750
Matches
1,343
Record
1017W – 326L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| SmolLM2-135M-Instruct-Q8_0 | 102.03 | 455.12 | 102.03 | 455.12 | 1 |
| gemma-3-270m-it-F16 | 40.84 | 202.25 | 40.84 | 202.25 | 1 |
| gemma-3-1b-it.Q5_K_M | 32.82 | 78.20 | 32.82 | 78.20 | 1 |
| llama-3.2-1b-instruct-q8_0 | 21.81 | 101.28 | 21.81 | 101.28 | 1 |
| qwen2.5-1.5b-instruct-q8_0 | 16.50 | 72.22 | 16.50 | 72.22 | 1 |
| SmolLM2-1.7B-Instruct-Q8_0 | 14.61 | 59.36 | 14.61 | 59.36 | 1 |
| qwen2.5-3b-instruct-q5_k_m | 12.81 | 22.94 | 12.81 | 22.94 | 1 |
| calme-3.3-instruct-3b-q5_0 | 12.56 | 21.27 | 12.56 | 21.27 | 1 |
| Gemmasutra-Mini-2B-v1-Q6_K | 12.23 | 32.71 | 12.23 | 32.71 | 1 |
| Qwen3-4B.Q4_K_M | 11.96 | 23.64 | 11.96 | 23.64 | 1 |
| gemma-2-2b-it-Q6_K | 11.53 | 32.07 | 12.68 | 34.58 | 3 |
| Phi-3.5-mini-instruct.Q4_K_M | 10.58 | 18.19 | 10.58 | 18.19 | 1 |
| gemma-3-4b-it.Q5_K_M | 10.36 | 19.73 | 10.36 | 19.73 | 1 |
| qwen2.5-coder-7b-instruct-q4_k_m | 7.13 | 17.43 | 7.13 | 17.43 | 1 |
| Qwen2.5-VL-7B-Instruct-Q4_K_M | 6.88 | 17.47 | 6.88 | 17.47 | 1 |
| DeepSeek-R1-Distill-Llama-8B-IQ4_XS | 6.60 | 8.21 | 6.60 | 8.21 | 1 |
| Llama-3.2-3B-Instruct-Q6_K | 6.46 | 22.34 | 7.02 | 27.52 | 2 |
| DeepSeek-R1-0528-Qwen3-8B-Q4_K_M | 6.16 | 13.52 | 6.16 | 13.52 | 1 |
| qwen3-7b-instruct-q4_k_m | 4.36 | 25.17 | 4.36 | 25.17 | 1 |
Head-to-Head Record
1–50 of 325 rows
1 / 7
Performance by App Version
ImprovedRegressed