Leaderboard

On-device LLM performance rankings powered by Glicko-2

iPhone 17

iOS

Rank

#12

Rating

1,954

±15 RD

Win Rate

93.8%

Conservative Rating

1,923

TG Rating

1,931

PP Rating

1,962

Matches

1,134

Record

1064W – 70L

Models Tested

ModelTG Median (tok/s)PP Median (tok/s)TG BestPP BestRuns
lille-130m-instruct-f3292.31374.7192.31374.711
gemma-3-1b-it-Q4_055.2595.1858.61100.832
Qwen3.5-0.8B-Q4_045.89757.1645.89757.161
llama-3.2-1b-instruct-q8_045.84773.3746.66826.283
LFM2-2.6B-Exp-Q4_K_M35.44296.2335.44296.231
Qwen3.5-2B.Q2_K33.23409.4333.23409.431
Gemmasutra-Mini-2B-v1-Q6_K26.52341.3126.52341.311
gemma-2-2b-it.Q6_K26.14345.0626.14345.061
gemma-2-2b-it-Q6_K25.29328.6727.22360.236
qwen2.5-3b-instruct-q5_k_m24.47231.5224.47231.521
gemma-3-4b-it-IQ4_NL23.07230.0123.07230.011
Huihui-Qwen3-4B-Thinking-2507-abliterated.Q4_K_M21.70187.4321.70187.431
Qwen3-VL-4B-Instruct-Q4_K_M21.69197.2421.69197.241
Qwen3.5-2B-Q8_021.41400.6621.41400.661
Qwen3-4B-Instruct-2507-UD-Q4_K_XL21.36190.1722.05204.642
Qwen3.5-2B.Q8_021.34411.9021.34411.901
Qwen.Qwen3-VL-Embedding-2B.Q8_017.98304.3017.98304.301
gemma-3-4b-it-q4_017.78210.1017.78210.101
gemma-3-4b-it-abliterated-v2.q6_k17.71226.4717.71226.471
guanaco-7b-uncensored.Q2_K17.19107.1117.33107.712
Llama-3.2-3B-Instruct-Q6_K16.22133.7621.64251.442
gemma-3n-E2B-it-Q5_K_S16.16159.1016.16159.101
qwen2.5-1.5b-instruct-q8_013.07248.6613.07248.661
SmolLM2-1.7B-Instruct-Q8_012.98211.1412.98211.141
Llama-3.2-3B-Instruct-Q4_012.5018.3912.5018.391
Qwen3.5-4B-IQ4_NL12.14142.6212.14142.621
Qwen3.5-4B-Q4_012.01153.2412.17153.922
Crow-4B-Opus-4.6-Distill-Heretic_Qwen3.5.i1-Q4_K_M11.36139.6411.36139.641
Qwen3.5-4B-Q4_110.22134.6110.22134.611
Phi-3.5-mini-instruct.Q4_K_M9.8714.829.8714.821
Qwen2.5-VL-7B-Instruct-iq2_m9.4776.439.4776.431

Head-to-Head Record

Performance by App Version

ImprovedRegressed

Compare With