Generative Question Answering on Bolmo Evaluation Suite GenQA 7B
81.6GenQA AverageLlama 3.1 70B
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Llama 3.1 70BModel Category=Open-weight, Number of Parameters=70B2025.12 | 81.6 | 88.4 | 91.7 | 79.6 | 92.4 | 78.3 | 84 | 53.1 | 92.9 | 73.9 | |
| Marin 32BModel Category=Fully-open, Number of Parameters=32B2025.12 | 80.3 | 87.2 | 90.5 | 76.7 | 91.1 | 76.5 | 80.5 | 55.1 | 94.4 | 70.7 | |
| Olmo 3 32BModel Category=Fully-open, Number of Parameters=32B2025.12 | 79.8 | 84.8 | 90.3 | 75.7 | 93.5 | 80.9 | 75.3 | 49 | 94.5 | 74.1 | |
| OLMo 2 32BModel Category=Fully-open, Number of Parameters=32B2025.12 | 79.1 | 87.5 | 89.4 | 77 | 88.7 | 76.3 | 79.1 | 51.4 | 94 | 68.7 | |
| Nemo. 3 Nano# Parameters=30B A3B, # Train Tokens=25T, Train Compute (10^23 FLOPs)=5.42026.04 | 78.1 | 86 | 88.7 | 75.9 | 92.5 | 75 | 77.2 | 45.7 | 94 | 67.7 | |
| Mistral 3.1 24BModel Category=Open-weight, Number of Parameters=24B2025.12 | 78 | 86.2 | 90.8 | 79.3 | 91.9 | 74.9 | 80.3 | 45.1 | 92.6 | 61.1 | |
| LLM 360 K2V2 70BModel Category=Fully-open, Number of Parameters=70B2025.12 | 77.1 | 87.6 | 88.9 | 76.8 | 90.6 | 69.7 | 79.8 | 47.6 | 91.2 | 61.5 | |
| Seed 36BModel Category=Open-weight, Number of Parameters=36B2025.12 | 76 | 84.8 | 89.3 | 76.1 | 96 | 76.1 | 77.4 | 30.7 | 89.1 | 64.4 | |
| Marin 8BOpenness=Fully-open, Parameters=8B2025.12 | 75.9 | 84 | 88.6 | 73.9 | 85.6 | 73 | 72.7 | 42.6 | 93.4 | 69.5 | |
| Kimi Linear# Parameters=48B A3B, # Train Tokens=6T, Train Compute (10^23 FLOPs)=1.02026.04 | 75.7 | 85 | 88.2 | 77.2 | 93.1 | 72.2 | 77.1 | 25.9 | 91.8 | 70.3 | |
| Gemma 2 9BOpenness=Open-weight, Parameters=9B2025.12 | 75.6 | 81.8 | 88.8 | 76.3 | 89.3 | 68.2 | 75.1 | 40.4 | 88.8 | 71.5 | |
| Apertus 70BModel Category=Fully-open, Number of Parameters=70B2025.12 | 75 | 84.5 | 87.7 | 74.8 | 87.5 | 56.3 | 77.2 | 43.1 | 90.7 | 72.8 | |
| Gemma 3 27BModel Category=Open-weight, Number of Parameters=27B2025.12 | 73.5 | 86 | 91.3 | 77.5 | 94.9 | 75.9 | 82.1 | 49.2 | 92.4 | 12.4 | |
| Llama 3.1 8BOpenness=Open-weight, Parameters=8B2025.12 | 73.1 | 81.5 | 87.3 | 75.5 | 88 | 59.5 | 70.9 | 36.7 | 89.2 | 69 | |
| Gemma 2 27BModel Category=Open-weight, Number of Parameters=27B2025.12 | 72.9 | 86.7 | 90.8 | 76.9 | 93.2 | 73.2 | 80.7 | 47.1 | 93 | 14.9 | |
| Olmo Hybrid# Parameters=7B, # Train Tokens=6T, Train Compute (10^23 FLOPs)=2.6, Decoding Settings=--disable-cascade-attn, --enforce-eager2026.04 | 72.9 | 79 | 86.2 | 70.2 | 89.7 | 72.9 | 64 | 34.8 | 92.1 | 67.4 | |
| Olmo 3 7BOpenness=Fully-open, Parameters=7B2025.12 | 72.5 | 77.7 | 85.7 | 68.9 | 89.5 | 71.5 | 60.4 | 32.6 | 93.5 | 72.8 | |
| Olmo 3# Parameters=7B, # Train Tokens=6T, Train Compute (10^23 FLOPs)=2.6, Decoding Settings=--disable-cascade-attn, --enforce-eager2026.04 | 72.5 | 77.7 | 85.6 | 68.8 | 89.5 | 71.5 | 60.3 | 32.5 | 93.6 | 72.8 | |
| OLMo 2 7BOpenness=Fully-open, Parameters=7B2025.12 | 72.4 | 82.2 | 87.4 | 70.5 | 82.2 | 61.5 | 70.8 | 37.4 | 91.5 | 68.3 | |
| Nemo-Nano 9BOpenness=Open-weight, Parameters=9B2025.12 | 71.8 | 80.2 | 86.2 | 67.9 | 91.4 | 71.4 | 64.9 | 31.2 | 92.3 | 60.4 | |
| Falcon H1# Parameters=7B, # Train Tokens=12T, Train Compute (10^23 FLOPs)=5.52026.04 | 71.7 | 80.9 | 87.9 | 73 | 93.3 | 69.4 | 61.1 | 27.4 | 91.4 | 60.7 | |
| MiMo 7BOpenness=Open-weight, Parameters=7B2025.12 | 71.4 | 80.6 | 86.5 | 73.1 | 89.7 | 69.3 | 65.6 | 33.1 | 90.3 | 54.4 | |
| Nemotron-H# Parameters=8B, # Train Tokens=15T, Train Compute (10^23 FLOPs)=7.22026.04 | 71.2 | 84.3 | 89.6 | 76 | 90.2 | 65.7 | 69.7 | 37.7 | 87.9 | 39.9 | |
| Qwen2 7BOpenness=Open-weight, Parameters=7B2025.12 | 71.1 | 80.5 | 86.4 | 73 | 93.5 | 57.2 | 65.1 | 33.8 | 89.2 | 61.6 | |
| Qwen 3# Parameters=8B, # Train Tokens=36T, Train Compute (10^23 FLOPs)=17.32026.04 | 71.1 | 80.6 | 86.4 | 72.8 | 93.4 | 57.2 | 65.1 | 33.8 | 89 | 61.8 | |
| Apertus 8BOpenness=Fully-open, Parameters=8B2025.12 | 69 | 81 | 85.8 | 70.9 | 83.8 | 37.1 | 70.1 | 35 | 89.6 | 67.4 | |
| Qwen 2.5 32BModel Category=Open-weight, Number of Parameters=32B2025.12 | 68.5 | 86.3 | 87.5 | 76.2 | 94.2 | 53.7 | 74 | 39.3 | 64.9 | 40.4 | |
| Recurr. Gemma# Parameters=9B, # Train Tokens=2T, Train Compute (10^23 FLOPs)=1.02026.04 | 68.5 | 81.3 | 86.3 | 72.2 | 83 | 44 | 66.4 | 30.6 | 85.9 | 66.6 | |
| Falcon Mamba# Parameters=7B, # Train Tokens=6T, Train Compute (10^23 FLOPs)=2.52026.04 | 68.5 | 81.9 | 88.3 | 80.9 | 86.1 | 41.6 | 67.1 | 35.1 | 75.9 | 59.7 | |
| Granite 3.3 8BOpenness=Open-weight, Parameters=8B2025.12 | 67.8 | 83.7 | 89.4 | 76 | 88.7 | 38.4 | 69.7 | 37 | 89.6 | 37.8 | |
| Qwen 2.5 7BOpenness=Open-weight, Parameters=7B2025.12 | 67.5 | 81 | 86 | 70.3 | 91.4 | 56.7 | 63 | 31.2 | 87 | 40.5 | |
| Gaperon 24BModel Category=Fully-open, Number of Parameters=24B2025.12 | 65.3 | 75.2 | 80.3 | 58.3 | 83.2 | 59.4 | 58.9 | 33.5 | 89.3 | 49.8 | |
| Gaperon 8BOpenness=Fully-open, Parameters=8B2025.12 | 63.3 | 73.9 | 76.4 | 67 | 80.5 | 54.9 | 55.5 | 28.8 | 86 | 46.7 | |
| xLSTM# Parameters=7B, # Train Tokens=2T, Train Compute (10^23 FLOPs)=0.92026.04 | 34.8 | 61.1 | 62.1 | 19.7 | 60.9 | 24.4 | 9.3 | 1.8 | 33.9 | 40.4 | |
| Olmo 3 7B# Parameters (incl. Embed)=7.30B2025.12 | 0.724 | 0.778 | 0.857 | 0.68 | 0.9 | 0.715 | 0.603 | 0.326 | 0.935 | 0.727 | |
| TFree-Hat 7B# Parameters (incl. Embed)=7.19B2025.12 | 0.713 | 0.828 | 0.882 | 0.705 | 0.896 | 0.486 | 0.683 | 0.343 | 0.886 | 0.711 | |
| Bolmo 7B# Parameters (incl. Embed)=7.63B2025.12 | 0.709 | 0.788 | 0.855 | 0.711 | 0.896 | 0.652 | 0.568 | 0.286 | 0.916 | 0.705 | |
| BLT 7B# Parameters (incl. Embed)=10.55B2025.12 | 0.684 | 0.811 | 0.882 | 0.728 | 0.845 | 0.388 | 0.673 | 0.292 | 0.852 | 0.686 | |
| EvaByte 6.5B# Parameters (incl. Embed)=6.49B2025.12 | 0.414 | 0.701 | 0.782 | 0.629 | 0.827 | 0.078 | 0.131 | 0.054 | 0.359 | 0.167 |