新浪开源模型VibeThinker-3B旨在展示推理可良好压缩但事实知识不行

Weibo's new VibeThinker-3B has just three billion parameters but matches top models up to 333 times its size on math and coding benchmarks. The performance comes from multi-stage post-training applied to an Alibaba base model. On tasks that require broad factual knowledge, however, the small model falls well behind. The researchers conclude that structured logical reasoning relies on few patterns and compresses well, while broad world knowledge still needs large models. A Chinese language model with just three billion parameters sometimes matches models a hundred times larger on math and coding tasks. The researchers behind it have developed a hypothesis about how AI capabilities are structured. Weibo's parent company Sina has released a small language model that competes with today's top models on hard math and coding tasks. According to a technical report , VibeThinker-3B performs on par with DeepSeek V3.2 and Kimi K2.5 on competitive benchmarks like AIME26. Both of those models have 200 to 333 times more parameters. Sina positions the model as an experiment in figuring out how much compute a model actually needs to compete at the top. Its predecessor, VibeThinker-1.5B, launched in November 2025. The new version pushes further, asking whether a small model can hit genuine top-tier performance, not just be "good for its size." Ad The results tell two different stories. On structured tasks with clearly verifiable solutions, like math olympiads or programming challenges, VibeThinker-3B matches models like GLM-5 or Gemini 3 Pro. On LiveCodeBench, it beats every other model under 20 billion parameters. Ad DEC_D_Incontent-1 Factual knowledge is a different story. On the knowledge-heavy GPQA-Diamond benchmark, the model falls well behind its much larger competitors. To rule out data contamination, the team had the model compete in LeetCode contests held between late April and late May 2026, after training wrapped up. VibeThinker-3B solved 123 out of 128 problems on the first try. That puts it ahead of GPT-5.2, Qwen3-Max, Kimi K2.5, and Claude Opus 4.6. It trails only GPT-5.3-Codex, Gemini 3.1 Pro, and Gemini 3 Flash, but not by much. Ad VibeThinker-3B builds on Alibaba's Qwen2.5-Coder-3B. Sina's contribution is the post-training, everything that happens after generic pre-training on large data sets. According to the report, that's what brings a 3B model close to the top performers. Post-training happens in stages. First, the model learns a broad range of tasks through supervised fine-tuning, covering math, coding, and general dialogue. Then it gets tailored for hard, multi-step reasoning problems. Ad DEC_D_Incontent-2 Reinforcement learning follows, applied sequentially for math, programming, and STEM. Self-distillation then consolidates the skills from each phase into a single model. A final step makes sure the model better follows instructions. Ad