Leaderboard
On-device LLM performance rankings powered by Glicko-2
Galaxy Z Fold7
AndroidRank
#42
Rating
1,815
±14 RD
Win Rate
80.3%
Conservative Rating
1,786
TG Rating
1,802
PP Rating
1,820
Matches
1,287
Record
1034W – 253L
Models Tested
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| tinygemma3-Q8_0 | 112.02 | 1129.13 | 112.02 | 1129.13 | 1 |
| Vintern-1B-v3_5-Q8_0 | 72.52 | 327.52 | 72.52 | 327.52 | 1 |
| Novaciano-The_Pervert-NSFW-RP-3.2-1B.i1-Q6_K | 44.78 | 133.74 | 44.78 | 133.74 | 1 |
| llama-3.2-1b-instruct-q8_0 | 36.74 | 134.94 | 36.74 | 134.94 | 1 |
| qwen2.5-1.5b-instruct-q8_0 | 24.22 | 456.40 | 27.09 | 521.06 | 2 |
| LFM2.5-1.2B-Thinking.i1-Q6_K | 20.55 | 60.67 | 20.55 | 60.67 | 1 |
| granite-4.0-h-tiny.i1-IQ4_NL | 15.97 | 64.68 | 15.97 | 64.68 | 1 |
| Qwen3-4B.Q4_K_M | 15.17 | 27.25 | 15.17 | 27.25 | 1 |
| nvidia_OpenReasoning-Nemotron-1.5B-IQ3_M | 15.03 | 22.34 | 15.03 | 22.34 | 1 |
| AI21-Jamba2-3B.i1-Q4_K_M | 14.20 | 44.90 | 14.20 | 44.90 | 1 |
| Phi-3.5-mini-instruct.Q4_K_M | 13.85 | 23.54 | 15.19 | 23.84 | 2 |
| gemma-2-2b-it-Q6_K | 11.48 | 40.51 | 13.52 | 101.13 | 3 |
| Gemmasutra-Mini-2B-v1-Q6_K | 10.88 | 46.89 | 16.46 | 59.44 | 3 |
| gemma-3-4b-it-Q4_K_M | 10.08 | 30.25 | 10.08 | 30.25 | 1 |
| Llama-3.2-3B-Instruct-Q6_K | 9.65 | 25.38 | 9.65 | 25.38 | 1 |
| AI21-Jamba2-3B.i1-Q6_K | 7.96 | 21.36 | 7.96 | 21.36 | 1 |
| qwen2.5-3b-instruct-q5_k_m | 7.26 | 63.91 | 7.26 | 63.91 | 1 |
| Dirty-Muse-Writer-v01-Uncensored-Erotica-NSFW.i1-Q4_K_M | 5.15 | 15.38 | 6.47 | 18.19 | 4 |
| gemma-2-9b-it-abliterated-Q4_K_M | 3.15 | 11.88 | 3.15 | 11.88 | 1 |
| Gemmasutra-Mini-2B-v1-f16 | 2.92 | 9.33 | 2.92 | 9.33 | 1 |
| gemma-3-12b-it-Q3_K_L | 2.23 | 3.79 | 2.23 | 3.79 | 1 |
| DeepSeek-R1-0528-Qwen3-8B-Q6_K | 2.15 | 7.14 | 2.15 | 7.14 | 1 |
Head-to-Head Record
1–50 of 313 rows
1 / 7
Performance by App Version
ImprovedRegressed