Leaderboard
On-device LLM performance rankings powered by Glicko-2
iQOO 13
AndroidRank
#41
Rating
1,841
±23 RD
Win Rate
83.0%
Conservative Rating
1,795
TG Rating
1,870
PP Rating
1,734
Matches
494
Record
410W – 84L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| llama-3.2-1b-instruct-q8_0 | 38.76 | 197.60 | 38.76 | 197.60 | 1 |
| DeepSeek-R1-Distill-Qwen-1.5B-Q3_K_XL | 35.31 | 63.75 | 36.64 | 66.63 | 2 |
| nvidia_OpenReasoning-Nemotron-1.5B-Q4_K_M | 32.80 | 60.64 | 32.80 | 60.64 | 1 |
| qwen2.5-1.5b-instruct-q8_0 | 30.60 | 119.32 | 30.60 | 119.32 | 1 |
| Phi-3.5-mini-instruct.Q4_K_M | 18.31 | 29.45 | 18.31 | 29.45 | 1 |
| Gemmasutra-Mini-2B-v1-Q6_K | 16.23 | 26.79 | 16.23 | 26.79 | 1 |
| qwen2.5-3b-instruct-q5_k_m | 14.49 | 22.47 | 18.24 | 29.28 | 2 |
| llava-v1.6-mistral-7b.Q8_0 | 6.93 | 15.16 | 6.93 | 15.16 | 1 |
Head-to-Head Record
1–50 of 208 rows
1 / 5
Performance by App Version
ImprovedRegressed