Leaderboard
On-device LLM performance rankings powered by Glicko-2
Nothing Phone (1)
AndroidRank
#160
Rating
1,420
±14 RD
Win Rate
42.3%
Conservative Rating
1,392
TG Rating
1,426
PP Rating
1,455
Matches
1,353
Record
572W – 781L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| SmolLM2-135M-Instruct-Q4_K_M | 43.34 | 193.51 | 43.34 | 193.51 | 1 |
| gemma-3-270m-it-F16 | 20.04 | 88.70 | 20.04 | 88.70 | 1 |
| LFM2.5-1.2B-Instruct-Q4_K_M | 15.44 | 60.87 | 15.44 | 60.87 | 1 |
| DeepSeek-R1-Distill-Qwen-1.5B-Q4_K_M | 9.79 | 36.37 | 9.79 | 36.37 | 1 |
| llama-3.2-1b-instruct-q8_0 | 8.29 | 47.70 | 8.29 | 47.70 | 1 |
| qwen2.5-1.5b-instruct-q8_0 | 6.83 | 36.47 | 7.17 | 36.71 | 2 |
| SmolLM2-1.7B-Instruct-Q8_0 | 6.20 | 24.22 | 6.20 | 24.22 | 1 |
| qwen2.5-3b-instruct-q5_k_m | 5.16 | 12.48 | 5.16 | 12.48 | 1 |
| gemma-2-2b-it-Q6_K | 4.92 | 15.28 | 5.12 | 15.69 | 5 |
| Phi-3.5-mini-instruct.Q4_K_M | 4.82 | 12.07 | 4.82 | 12.07 | 1 |
| Gemmasutra-Mini-2B-v1-Q6_K | 4.12 | 11.45 | 5.01 | 14.01 | 2 |
| Qwen3-4B-Instruct-2507-IQ2_S-2.34bpw | 3.72 | 7.07 | 3.72 | 7.07 | 1 |
| gemma-3-4b-it-q4_0 | 3.00 | 15.85 | 3.00 | 15.85 | 1 |
| DeepSeek-R1-Distill-Qwen-7B-Q4_K_M | 2.23 | 7.20 | 2.23 | 7.20 | 1 |
| Llama-3.2-3B-Instruct-Q6_K | 2.02 | 6.40 | 3.83 | 10.32 | 2 |
| gemma-3n-E4B-it-Q4_K_M | 1.94 | 6.13 | 1.94 | 6.13 | 1 |
Head-to-Head Record
1–50 of 324 rows
1 / 7
Performance by App Version
ImprovedRegressed