Solar Open 2 发了，250B MoE 模型，1M token 超长上下文，专门跑复杂智能体任务。MMLU-Pro 和 LiveCodeBench 都领先，韩语能力还特别强，比 DeepSeek-V4-Pro 小很多但性能接近。
Solar Open 2 是一个 250B 参数（15B 激活）的混合专家（MoE）语言模型，上下文窗口达 1M token。它采用混合注意力机制，每 3 层线性注意力插入 1 层 softmax，无位置编码。训练上，从 Solar Open 1 初始化 5.69B 参数共享骨架，并通过质量与稀有度感知的数据筛选将 20T 语料缩减为 10T 以提升效率。通过多教师同策略蒸馏（MOPD）融合 12 个领域专家。在 MMLU-Pro、LiveCodeBench 和 APEX-Agents 上领先，韩语基准平均最高，Ko-GDPval 上以不到 1/6 参数量与 DeepSeek-V4-Pro (1.6T) 竞争。
Solar Open 2 Technical Report
We present Solar Open 2, a 250B-A15B Mixture-of-Experts language model built for long-horizon agentic tasks, scaled up from Solar Open 1 (Solar Open 100B). To hold entire agent trajectories in a single context, Solar Open 2 reaches a 1M-token window through a hybrid attention stack that interleaves one softmax layer among every three linear-attention layers, using no positional encoding and a gated delta rule extended to negative eigenvalues. To train at this scale under a fixed compute budget, we make training efficient in two ways: a stronger starting point, and higher-value data. For the starting point, we initialize Solar Open 2 from Solar Open 1, transferring the 5.69B-parameter shared skeleton that survives the architectural change and learning everything else through full pre-training. For the data, we curate for value per token: quality- and rarity-aware data curation and mixture-ratio optimization refine a 20T pool into a 10T mixture that, at equal token budget, outperforms the Solar Open 1 recipe. To build its agent skills, we train twelve domain specialists across purpose-built scenarios, then consolidate them into a single model by Multi-teacher On-Policy Distillation (MOPD). Against comparably sized open-weight models on English benchmarks, Solar Open 2 leads on MMLU-Pro, LiveCodeBench, and the APEX-Agents agentic suite, and stays competitive with the strongest (DeepSeek-V4-Flash and MiMo-V2.5) elsewhere. On Korean benchmarks, Solar Open 2 records the highest average of any model compared, including fast-tier closed APIs, and on Ko-GDPval, an in-house Korean officework-agent benchmark, it is competitive with DeepSeek-V4-Pro (1.6T) at less than a sixth of its size.