Leaderboard
On-device LLM performance rankings powered by Glicko-2
Galaxy A34 5G
AndroidRank
#232
Rating
1,150
±19 RD
Win Rate
16.2%
Conservative Rating
1,111
TG Rating
1,159
PP Rating
1,111
Matches
716
Record
116W – 600L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| ggml-model-Q3_K_L | 14.10 | 52.45 | 14.10 | 52.45 | 1 |
| Llama-3.2-1B-Instruct-abliterated.i1-Q4_K_M | 10.87 | 37.92 | 10.87 | 37.92 | 1 |
| LFM2-VL-1.6B-Q4_0 | 9.56 | 58.30 | 9.56 | 58.30 | 1 |
| gemma-3-1b-it-Q4_K_M | 8.87 | 28.99 | 8.87 | 28.99 | 1 |
| gemma-3-1b-it.Q4_K_M | 8.77 | 28.36 | 9.00 | 28.80 | 5 |
| llama-3.2-1b-instruct-q8_0 | 6.30 | 48.38 | 6.30 | 48.38 | 1 |
| Dirty-Alice-RP-2b.i1-Q3_K_S | 4.62 | 9.36 | 4.62 | 9.36 | 1 |
| qwen2.5-3b-instruct-q5_k_m | 3.42 | 8.19 | 3.42 | 8.19 | 1 |
| Phi-3.5-mini-instruct.Q4_K_M | 3.29 | 6.51 | 3.96 | 6.56 | 2 |
| EXAONE-3.5-2.4B-Instruct-Q8_0 | 3.01 | 9.98 | 3.01 | 9.98 | 1 |
| DeepSeek-R1-Distill-Qwen-1.5B-Q4_K_M | 2.86 | 17.75 | 2.91 | 18.40 | 2 |
| Llama-3.2-3B-Instruct-Q6_K | 2.62 | 6.60 | 2.62 | 6.60 | 1 |
| qwen2.5-1.5b-instruct-q8_0 | 1.58 | 6.71 | 1.58 | 6.71 | 1 |
| Mistral-7B-Instruct-v0.3.IQ1_M | 1.38 | 1.79 | 1.38 | 1.79 | 1 |
Head-to-Head Record
1–50 of 238 rows
1 / 5
Performance by App Version
ImprovedRegressed