Leaderboard
On-device LLM performance rankings powered by Glicko-2
Loading device data...
On-device LLM performance rankings powered by Glicko-2
Loading device data...
Rank
#53
Rating
1,790
±19 RD
Win Rate
78.1%
Conservative Rating
1,752
TG Rating
1,820
PP Rating
1,748
Matches
729
Record
569W – 160L
| Model | TG Median (tok/s) | PP Median (tok/s) | TG Best | PP Best | Runs |
|---|---|---|---|---|---|
| gemma-3-1b-it-Q8_0 | 29.05 | 138.10 | 29.05 | 138.10 | 1 |
| llama-3.2-1b-instruct-q8_0 | 25.25 | 96.95 | 25.25 | 96.95 | 1 |
| qwen2.5-1.5b-instruct-q8_0 | 19.35 | 73.25 | 19.35 | 73.25 | 1 |
| Llama-3.2-3B-Instruct-Q4_0 | 16.02 | 63.37 | 16.02 | 63.37 | 1 |
| Phi-3.5-mini-instruct.Q4_K_M | 12.02 | 18.84 | 12.02 | 18.84 | 1 |
| Llama-3.2-3B-Instruct-Q6_K | 7.72 | 19.64 | 7.72 | 19.64 | 1 |
| deepseek-ai_DeepSeek-R1-0528-Qwen3-8B-Q4_0 | 7.53 | 27.11 | 7.53 | 27.11 | 1 |
| Phi-4-mini-instruct-Q4_K_M | 6.92 | 16.06 | 6.92 | 16.06 | 1 |
1–50 of 189 rows