Leaderboard
On-device LLM performance rankings powered by Glicko-2
Nothing Phone (2a)
AndroidRank
#184
Rating
1,335
±17 RD
Win Rate
34.1%
Conservative Rating
1,302
TG Rating
1,316
PP Rating
1,443
Matches
986
Record
336W – 650L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| zephyr-smol_llama-100m-sft-full.q2_k | 106.31 | 462.85 | 106.31 | 462.85 | 1 |
| Falcon-H1R-Tiny-90M-Q5_K_M | 51.52 | 178.32 | 51.52 | 178.32 | 1 |
| chatgpt-5-q8_0 | 13.15 | 79.73 | 13.15 | 79.73 | 1 |
| DeepSeek-R1-Distill-Qwen-1.5B-IQ2_M | 9.25 | 18.09 | 9.25 | 18.09 | 1 |
| Qwen3-Embedding-0.6B-Q8_0 | 8.32 | 81.65 | 8.32 | 81.65 | 1 |
| llama-3.2-1b-instruct-q8_0 | 6.69 | 56.22 | 8.06 | 58.29 | 3 |
| qwen2.5-1.5b-instruct-q8_0 | 5.42 | 34.52 | 5.42 | 34.52 | 1 |
| jan-nano-128k-Q4_K_M | 4.66 | 12.13 | 4.66 | 12.13 | 1 |
| qwen2.5-3b-instruct-q5_k_m | 4.14 | 13.02 | 4.14 | 13.02 | 1 |
| SmolLM2-1.7B-Instruct-Q8_0 | 3.98 | 34.48 | 3.98 | 34.48 | 1 |
| Llama-3.2-3B-Instruct-Q6_K | 2.97 | 10.06 | 3.68 | 11.42 | 2 |
| gemma-2-2b-it-Q6_K | 2.31 | 9.82 | 3.20 | 12.85 | 2 |
| Dolphin3.0-Llama3.1-8B-Q4_0 | 1.61 | 11.47 | 1.61 | 11.47 | 1 |
Head-to-Head Record
1–50 of 259 rows
1 / 6
Performance by App Version
ImprovedRegressed