3B (dense)
Apache 2.0
Compact reasoning model built on Nanbeige4-3B-Base with further SFT and RL post-training; pretrained on 23T high-quality tokens and finetuned on 30M+ diverse instructions; 64K context via RoPE with ABF technique; strong multi-step reasoning rivaling 30B models like Qwen3-32B; first general small model to natively support deep-search tasks with 500+ rounds of tool invocations; Fine-Grained Warmup-Stable-Decay (FG-WSD) training scheduler; Dual Preference Distillation (DPD) method; strong benchmarks: AIME 2026 I (87.4%), LiveCodeBench-Pro Easy (81.4%), Arena-Hard-v2 (73.2%), IMO-Answer-Bench; outperforms Qwen3-4B, Qwen3-8B, Qwen3-30B-A3B, Qwen3-32B on Arena-Hard-v2 and Multi-Challenge; supports chat, tool use, and deep-search scenarios