The typical standard errors between pairs of models on this dataset as a function of the absolute accuracy.
| model | pass1 | win_rate | count | SE(A) | SE_x(A) | SE_pred(A) |
|---|---|---|---|---|---|---|
| qwen3-14b | 72.9 | 35.9 | 3 | 2 | 1.8 | 0.79 |
| google_gemma_3_12b_it | 69.5 | 33.1 | 4 | 2.1 | 1.9 | 0.8 |
| qwen2.5-coder-14b-instruct | 68.1 | 34 | 3 | 2.1 | 1.6 | 1.3 |
| qwen3-8b | 65.3 | 30.2 | 3 | 2.1 | 1.9 | 1 |
| deepseek_r1_distill_llama_70b | 64.1 | 28.9 | 2 | 2.1 | 1.7 | 1.3 |
| qwen3-4b | 64 | 29.1 | 4 | 2.1 | 1.9 | 0.9 |
| google_gemma_3_4b_it | 60.3 | 26.3 | 5 | 2.2 | 2 | 0.89 |
| mistralai_mixtral_8x22b_instruct_v0.1 | 60.1 | 26.4 | 3 | 2.2 | 1.7 | 1.4 |
| qwen2.5-coder-7b-instruct | 57.3 | 26.7 | 4 | 2.2 | 1.5 | 1.7 |
| deepseek_r1_distill_qwen_14b | 53.3 | 22.1 | 4 | 2.2 | 1.7 | 1.4 |
| llama-3.1-8B-instruct | 52.2 | 21 | 7 | 2.2 | 2.2 | 0 |
| mistralai_mixtral_8x7b_instruct_v0.1 | 49.1 | 19.3 | 3 | 2.2 | 1.8 | 1.3 |
| qwen2-7b-instruct | 48 | 18.7 | 4 | 2.2 | 1.7 | 1.4 |
| qwen2.5-coder-3b-instruct | 47.9 | 19.9 | 4 | 2.2 | 1.5 | 1.6 |
| qwen3-1.7b | 47.8 | 18.5 | 4 | 2.2 | 1.9 | 1.1 |
| llama-3.2-3B-instruct | 45.8 | 17.3 | 8 | 2.2 | 2.2 | 0 |
| mistralai_ministral_8b_instruct_2410 | 44.9 | 16.7 | 4 | 2.2 | 1.7 | 1.5 |
| mistralai_mathstral_7b_v0.1 | 41.1 | 14.7 | 4 | 2.2 | 1.6 | 1.5 |
| deepseek_v2_lite_chat | 40.3 | 14.6 | 3 | 2.2 | 1.7 | 1.4 |
| qwen1.5-14b-chat | 37.2 | 12.3 | 3 | 2.2 | 1.7 | 1.3 |
| mistralai_mistral_7b_instruct_v0.3 | 36.8 | 12.1 | 4 | 2.2 | 1.7 | 1.3 |
| deepseek_r1_distill_llama_8b | 34.2 | 11.8 | 4 | 2.1 | 1.5 | 1.5 |
| mistralai_mistral_7b_instruct_v0.2 | 34.1 | 11.3 | 4 | 2.1 | 1.6 | 1.4 |
| qwen2.5-coder-1.5b-instruct | 32.8 | 11.3 | 4 | 2.1 | 1.3 | 1.6 |
| deepseek_r1_distill_qwen_7b | 32.3 | 11.5 | 4 | 2.1 | 1.4 | 1.6 |
| qwen1.5-7b-chat | 31.9 | 9.96 | 3 | 2.1 | 1.6 | 1.3 |
| llama-3.2-1B-instruct | 26.2 | 8.06 | 11 | 2 | 2 | 0 |
| qwen2.5-coder-0.5b-instruct | 25 | 7.9 | 5 | 1.9 | 1.3 | 1.5 |
| qwen3-0.6b | 24.1 | 6.9 | 5 | 1.9 | 1.5 | 1.2 |
| mistralai_mistral_7b_instruct_v0.1 | 22.9 | 6.63 | 4 | 1.9 | 1.2 | 1.4 |
| qwen2-1.5b-instruct | 12.4 | 3 | 4 | 1.5 | 0.9 | 1.2 |
| deepseek_r1_distill_qwen_1.5b | 11.2 | 2.86 | 4 | 1.4 | 0.88 | 1.1 |
| qwen1.5-1.8b-chat | 7.13 | 1.58 | 3 | 1.2 | 0.7 | 0.92 |
| qwen2-0.5b-instruct | 6.72 | 1.44 | 5 | 1.1 | 0.59 | 0.95 |
| qwen1.5-0.5b-chat | 1.12 | 0.179 | 5 | 0.47 | 0.18 | 0.43 |