Filipino Performance
Filipino Scores by Model
Average of 2000 bootstraps. 95% CI are shown.
Model Size: ≤200B
Open instruct models only
Filipino Competency & Task Scores
Average of 2000 bootstraps. 95% CI are shown.
Model Size: ≤200B
Open instruct models only
Model | TL | Cultural | Inst. Following | Knowledge | Multi-Turn Chat | NLG | NLR | NLU | Safety |
|---|---|---|---|---|---|---|---|---|---|
 Gemma 4 31B | 77.12 [-1.48, +1.53] | 78.95 [-4.25, +4.09] | 92.13 [-5.10, +4.66] | 75.74 [-4.74, +4.67] | 82.83 [-4.14, +3.85] | 58.72 [-0.90, +0.91] | 81.44 [-2.85, +2.88] | 75.26 [-5.11, +4.76] | 71.89 [-4.94, +4.91] |
 SEA-LION v4.5 (Qwen) 27B AISG | 73.28 [-1.53, +1.52] | 74.45 [-4.18, +4.13] | 87.77 [-5.50, +5.09] | 73.38 [-4.58, +4.87] | 80.01 [-4.07, +3.76] | 55.90 [-0.74, +0.76] | 78.91 [-2.94, +2.86] | 71.10 [-5.02, +4.86] | 64.74 [-5.87, +5.47] |
 Qwen 3.5 122B MoE Alibaba | 72.33 [-1.40, +1.41] | 74.03 [-4.42, +3.98] | 89.77 [-4.88, +4.40] | 73.85 [-4.61, +4.65] | 81.87 [-3.77, +3.74] | 55.44 [-0.65, +0.66] | 73.00 [-3.54, +3.15] | 70.00 [-5.07, +4.75] | 60.68 [-4.45, +4.51] |
Qwen 3.6 27B Alibaba | 67.43 [-1.43, +1.41] | 57.37 [-4.25, +4.18] | 88.94 [-5.25, +4.63] | 69.67 [-4.47, +4.70] | 80.28 [-4.00, +3.85] | 53.44 [-0.65, +0.66] | 68.37 [-3.06, +2.97] | 63.76 [-4.58, +4.44] | 57.62 [-4.14, +4.23] |
 Mistral Medium 3.5 128B Mistral AI | 66.74 [-1.57, +1.63] | 67.29 [-4.27, +4.01] | 74.94 [-6.72, +6.26] | 67.18 [-5.35, +4.78] | 78.93 [-4.23, +4.06] | 54.23 [-0.62, +0.64] | 63.64 [-3.42, +3.46] | 68.96 [-4.61, +4.38] | 58.74 [-5.26, +5.20] |
 Llama 4 Scout 109B MoE Meta | 65.28 [-1.67, +1.68] | 65.35 [-4.33, +4.43] | 89.72 [-5.31, +4.57] | 68.23 [-5.31, +5.31] | 63.96 [-5.21, +5.62] | 55.08 [-0.64, +0.65] | 68.59 [-3.64, +3.65] | 66.46 [-5.19, +4.93] | 44.84 [-5.85, +6.20] |
 Nemotron 3 Super 120B MoE NVIDIA | 55.51 [-1.51, +1.57] | 61.97 [-3.78, +3.66] | 66.98 [-7.82, +7.42] | 58.84 [-4.93, +5.03] | 79.90 [-3.88, +3.78] | 47.56 [-0.52, +0.58] | 49.06 [-3.23, +2.96] | 65.50 [-4.79, +4.40] | 14.29 [-2.82, +2.74] |
 MERaLiON 2 10B A*STAR | 54.24 [-1.69, +1.66] | 52.41 [-3.72, +3.99] | 56.31 [-7.50, +7.51] | 49.59 [-6.22, +5.95] | 44.16 [-4.49, +4.54] | 51.11 [-0.59, +0.59] | 64.32 [-3.70, +3.39] | 66.76 [-5.69, +5.27] | 49.24 [-4.34, +4.52] |
SEA-LION v4.5 (Gemma E2B) 5B MoE AISG | 53.04 [-1.87, +1.87] | 48.23 [-4.81, +4.80] | 72.32 [-7.20, +6.85] | 36.38 [-6.38, +6.41] | 68.43 [-5.02, +4.99] | 54.38 [-0.71, +0.72] | 47.23 [-4.78, +4.76] | 60.59 [-5.94, +5.92] | 36.74 [-5.05, +5.15] |
Gemma 4 (E2B) 5B | 51.63 [-1.96, +2.00] | 41.04 [-4.72, +4.81] | 72.34 [-6.99, +6.70] | 35.28 [-6.23, +6.18] | 70.87 [-4.96, +4.99] | 54.03 [-0.69, +0.68] | 40.69 [-4.86, +4.68] | 61.10 [-5.92, +5.58] | 37.67 [-5.99, +6.35] |
 Olmo 3.1 32B AI2 | 48.91 [-1.92, +1.87] | 40.94 [-4.39, +4.11] | 63.96 [-8.01, +8.19] | 40.76 [-6.26, +5.78] | 66.22 [-6.29, +6.02] | 49.62 [-0.65, +0.67] | 40.64 [-4.43, +4.44] | 60.34 [-6.16, +5.76] | 28.77 [-4.69, +4.59] |
 GLM 4.7 Flash 30B MoE Z.ai | 39.92 [-1.50, +1.55] | 31.31 [-3.15, +2.97] | 58.00 [-6.45, +6.29] | 22.87 [-3.79, +3.92] | 42.34 [-5.21, +5.53] | 47.57 [-0.65, +0.68] | 23.29 [-2.44, +2.45] | 55.71 [-4.87, +4.32] | 38.25 [-5.04, +4.80] |
 Apertus 8B Swiss AI | 24.15 [-1.35, +1.36] | 15.14 [-2.66, +2.87] | 56.16 [-6.99, +6.58] | 0.07 [-0.07, +0.97] | 32.34 [-5.22, +5.41] | 42.99 [-0.54, +0.54] | 5.72 [-1.46, +1.59] | 26.01 [-4.74, +5.04] | 14.79 [-3.07, +3.31] |
 Tiny Aya Water 3B CohereLabs | 24.05 [-1.30, +1.19] | 11.98 [-2.35, +2.52] | 67.03 [-6.31, +5.71] | 22.71 [-4.25, +4.67] | 32.37 [-4.43, +4.46] | 36.14 [-0.55, +0.56] | 2.26 [-2.19, +2.08] | 6.52 [-3.10, +3.09] | 13.37 [-2.29, +2.37] |