Leaderboard
On-device LLM performance rankings powered by Glicko-2
Galaxy S23 FE
AndroidRank
#190
Rating
1,314
±16 RD
Win Rate
32.1%
Conservative Rating
1,282
TG Rating
1,311
PP Rating
1,419
Matches
1,044
Record
335W – 709L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| tinygemma3-Q8_0 | 33.07 | 587.87 | 33.07 | 587.87 | 1 |
| LFM2-VL-1.6B-Q8_0 | 15.05 | 67.59 | 15.05 | 67.59 | 1 |
| gemma-3-1b-it-Q4_0 | 13.49 | 79.15 | 13.49 | 79.15 | 1 |
| LiquidAI_LFM2-VL-1.6B-IQ4_NL | 12.03 | 91.05 | 12.03 | 91.05 | 1 |
| llama-3.2-1b-instruct-q8_0 | 8.53 | 56.21 | 8.53 | 56.21 | 1 |
| LFM2-2.6B-Q4_0 | 8.04 | 46.25 | 8.04 | 46.25 | 1 |
| Phi-3.5-mini-instruct.Q4_K_M | 6.04 | 12.30 | 6.04 | 12.30 | 1 |
| SmolLM2-1.7B-Instruct-Q8_0 | 5.99 | 33.68 | 5.99 | 33.68 | 1 |
| gemma-2-2b-it-Q6_K | 5.66 | 15.33 | 5.66 | 15.33 | 1 |
| LFM2-2.6B-Q6_K | 5.43 | 16.40 | 5.43 | 16.40 | 1 |
| Qwen3-4B-Instruct-2507-Q4_0 | 5.20 | 25.47 | 5.20 | 25.47 | 1 |
| qwen2.5-1.5b-instruct-q8_0 | 3.22 | 30.23 | 3.22 | 30.23 | 1 |
| Llama-3.2-3B-Instruct-Q6_K | 3.15 | 7.96 | 3.15 | 7.96 | 1 |
| qwen2.5-3b-instruct-q5_k_m | 2.84 | 11.53 | 2.84 | 11.53 | 1 |
Head-to-Head Record
1–50 of 301 rows
1 / 7