Leaderboard
On-device LLM performance rankings powered by Glicko-2
Group by:ModelDevicePlatform
Showing 6,724 submissions → 4,651 rows
| Device | Platform | PP Avg (t/s) | TG Avg (t/s) | Submissions | Model Name | PP Config | TG Config | Flash Attn |
|---|---|---|---|---|---|---|---|---|
| iPhone 12 Pro | iOS | 237967.5 | 4115.3 | 1 | DeepSeek-R1-Distill-Qwen-1.5B-BF16. | 512 | 128 | ✗ |
| iPhone 15 Pro | iOS | 116609.3 | 2442.1 | 1 | EXAONE-3.5-7.8B-Instruct-Q4_K_M. | 512 | 128 | ✗ |
| iPhone 11 Pro | iOS | 226935.1 | 2409.5 | 1 | DeepSeek-R1-Distill-Qwen-1.5B-Q8_0. | 512 | 128 | ✗ |
| iPhone 14 | iOS | 304851.4 | 2253.9 | 4 | DeepSeek-R1-Distill-Qwen-1.5B-Q8_0. | 512 | 128 | ✗ |
| iPhone 15 Pro Max | iOS | 94180.1 | 2113.6 | 1 | llama-3.1-8b-instruct-q4_0. | 512 | 128 | ✗ |
| iPad Air 5th Gen | iOS | 55308.4 | 1683.7 | 1 | Qwen3-8B-Q4_K_M. | 128 | 256 | ✗ |
| Galaxy S25 | Android | 116475.3 | 1564.3 | 1 | tinystories-lay1-hs128-hd2-262K.Q4_0. | 512 | 128 | ✗ |
| iPad Pro 12.9 inch 6th Gen | iOS | 87452.5 | 1542.9 | 1 | DeepSeek-R1-0528-Qwen3-8B-Q4_K_M. | 512 | 128 | ✗ |
| iPhone 16 Pro Max | iOS | 78705.3 | 1385.5 | 2 | DeepSeek-R1-Distill-Qwen-7B-Q4_K_M. | 512 | 128 | ✗ |
| iPhone 13 Pro Max | iOS | 42173.9 | 1359.5 | 1 | yi-ai-4b-chinese-it-v1-q6_k. | 512 | 128 | ✗ |
| iPhone 15 Pro | iOS | 120381.8 | 1336.5 | 1 | Qwen3-8B-Q4_K_M. | 512 | 128 | ✗ |
| iPhone 15 Pro Max | iOS | 37109.4 | 1179.5 | 1 | DeepSeek-R1-Distill-Qwen-14B-IQ2_XS. | 128 | 32 | ✗ |
| POCO F8 Ultra | Android | 67352.7 | 974.7 | 1 | tinystories-lay1-hs128-hd2-262K.Q4_0. | 512 | 128 | ✗ |
| Galaxy S25 Ultra | Android | 2095.9 | 477.2 | 1 | FedPPO-Confused-Pythia-70M-a0.Q2_K. | 128 | 32 | ✗ |
| iPhone 13 Pro | iOS | 6075.1 | 450.1 | 3 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| iPhone 14 Pro Max | iOS | 23733.4 | 394.3 | 2 | DeepSeek-R1-Distill-Qwen-7B-IQ2_M. | 128 | 32 | ✗ |
| iPhone 16 Pro | iOS | 6180.9 | 387.4 | 2 | Phi-4-mini-instruct-Q4_K_M. | 512 | 128 | ✗ |
| iPhone 14 Pro | iOS | 935.7 | 352.4 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| Galaxy S25 | Android | 1245.1 | 352.0 | 1 | SmolLM2-135M-Instruct-Q4_0. | 512 | 128 | ✗ |
| iPhone 16 Pro Max | iOS | 25507.5 | 324.4 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| Galaxy Z Fold6 | Android | 1572.9 | 313.8 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| iPhone 15 Pro Max | iOS | 20905.9 | 284.2 | 1 | DeepSeek-R1-Distill-Qwen-14B-IQ2_S. | 512 | 128 | ✗ |
| Galaxy S24 | Android | 1435.0 | 276.4 | 1 | FuadeAI-50M.Q4_K_S. | 512 | 128 | ✗ |
| Galaxy S25 Ultra | Android | 13650.8 | 251.8 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| Galaxy S25 | Android | 954.2 | 244.6 | 1 | SmolLM2-135M-Instruct-Q8_0. | 512 | 128 | ✗ |
| Galaxy S24+ | Android | 1134.4 | 237.7 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| Galaxy S24 | Android | 771.6 | 235.6 | 1 | zephyr-smol_llama-100m-sft-full.q3_k_m. | 512 | 128 | ✗ |
| iPad Air 11-inch 7th Gen | iOS | 5861.3 | 224.9 | 2 | SmolLM2-135M-Instruct-Q2_K. | 512 | 2048 | ✗ |
| iPad Pro 11 inch 5th Gen | iOS | 5714.4 | 206.5 | 2 | SmolLM2-135M-Instruct-Q8_0. | 512 | 128 | ✗ |
| iPhone 16 Pro | iOS | 4223.2 | 193.3 | 2 | SmolLM2-135M-Instruct-Q4_0. | 512 | 128 | ✗ |
| iPhone 16 | iOS | 2897.7 | 181.2 | 1 | google_gemma-3n-E2B-it-Q8_0. | 512 | 128 | ✗ |
| iPhone 16 Pro Max | iOS | 5002.2 | 177.2 | 2 | SmolLM2-135M-Instruct-Q8_0. | 512 | 128 | ✗ |
| iPhone 13 Pro | iOS | 419.0 | 176.8 | 10 | SmolLM2-135M-Instruct-Q4_0. | 64 | 32 | ✗ |
| iPhone 16 Pro | iOS | 2378.8 | 171.8 | 1 | gemma-3n-E4B-it-IQ4_NL. | 512 | 128 | ✗ |
| Mate 20 Pro | Android | 850.3 | 167.7 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| iPhone 16 | iOS | 3208.2 | 166.4 | 1 | gemma-3n-E2B-it-Q8_0. | 512 | 128 | ✗ |
| iPhone 16 Pro | iOS | 4162.8 | 164.9 | 2 | SmolLM2-135M-Instruct-Q8_0. | 512 | 128 | ✗ |
| Xiaomi 14T Pro | Android | 1135.4 | 164.1 | 1 | zephyr-smol_llama-100m-sft-full-Q8_0. | 64 | 32 | ✗ |
| WP30_Pro_EEA | Android | 917.2 | 163.3 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| ZenFone 7 / 7 Pro (ZS670KS/ZS671KS) | Android | 749.6 | 159.6 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| iPhone 7 Plus | iOS | 869.4 | 158.4 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| iPhone 16 Plus | iOS | 3139.6 | 154.4 | 3 | SmolLM2-135M-Instruct-Q4_0. | 512 | 128 | ✗ |
| POCO F8 Ultra | Android | 746.0 | 148.5 | 1 | Falcon-H1-Tiny-R-90M.Q8_0. | 512 | 128 | ✗ |
| iPhone 14 Pro Max | iOS | 22221.2 | 143.7 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| Pixel 7a | Android | 1101.3 | 138.8 | 1 | tinygemma3-Q8_0. | 512 | 128 | ✗ |
| iPhone 16 Pro | iOS | 4139.7 | 130.9 | 1 | google_gemma-3-270m-it-Q8_0. | 512 | 128 | ✗ |
| iPhone 12 | iOS | 276.0 | 126.0 | 1 | SmolVLM-256M-Instruct-Q8_0. | 512 | 128 | ✗ |
| iPhone 13 Pro | iOS | 239.5 | 123.9 | 1 | SmolLM2-135M-Instruct-IQ3_M. | 512 | 128 | ✗ |
| iPad Pro 11 inch 4th Gen | iOS | 1162.4 | 122.5 | 1 | Meta-Llama-3-8B-Instruct.Q4_K_M. | 512 | 128 | ✗ |
| iPhone 15 Pro | iOS | 303.4 | 121.3 | 1 | gemma-3-270m-it-Q2_K. | 512 | 128 | ✗ |
1–50 of 4,651 rows
1 / 94