Leaderboard

On-device LLM performance rankings powered by Glicko-2

iPad Pro 12.9 inch 7th Gen

iOS

Rank

#2

Rating

2,013

±16 RD

Win Rate

99.5%

Conservative Rating

1,980

TG Rating

2,011

PP Rating

2,013

Matches

1,005

Record

1000W – 5L

Models Tested

ModelTG Median (tok/s)PP Median (tok/s)TG BestPP BestRuns
gemma-3-1b-it.Q4_K_S85.61181.2485.61181.241
llama-3.2-1b-instruct-q8_058.901350.5258.901350.521
gemma-2-2b-it-Q6_K40.96589.7642.40663.203
qwen2.5-3b-instruct-q5_k_m36.93424.8736.93424.871
Qwen3-4B-UD-Q4_K_XL34.96346.0534.96346.051
gemma-2-2b-it.Q5_K_M30.49551.9931.86556.032
Yi-Coder-1.5B-Chat.fp1630.4864.2130.4864.211
Phi-3.5-mini-instruct.Q4_K_M30.05196.0235.98358.142
qwen2.5-3b-instruct-q8_028.9962.9328.9962.931
Llama-3.2-3B-Instruct-Q6_K28.38376.2034.43479.356
Qwen3-8B.Q3_K_M18.97166.1218.97166.121
ai21labs_AI21-Jamba-Reasoning-3B-Q8_017.7843.1617.7843.161
DeepSeek-R1-0528-Qwen3-8B-IQ4_NL11.2215.6511.2215.651
DeepSeek-R1-Distill-Qwen-7B-Q4_K_L2.4110.632.4110.631
Llama-3.2-9B-Uncensored-Brainstorm-Alpha-D_AU-IQ4_XS0.2313.140.2313.141

Head-to-Head Record

Performance by App Version

ImprovedRegressed

Compare With