VibeThinker-3B: Exploring the Frontier of Verifiable Reasoning in Small Language Models
This technical report introduces VibeThinker-3B, a compact dense model with 3B parameters developed to investigate ho...
Source Evidence
What Changed
This technical report introduces VibeThinker-3B, a compact dense model with 3B parameters developed to investigate ho...
Why It Matters
VibeThinker‑3B shows that verifiable reasoning can be distilled into a 3 B‑parameter core, eroding the long‑standing assumption that only >10 B models can pass complex reasoning benchmarks—this forces a re‑examination of deployment economics and benchmark design. Its success also fuels the Parametric Compression‑Coverage Hypothesis, which could shift future research toward building highly compressed reasoning engines while keeping larger models for open‑domain knowledge, reshaping the competitive landscape for both cloud and edge AI services.
Confirmed Facts
This technical report introduces VibeThinker-3B, a compact dense model with 3B parameters developed to investigate how far verifiable reasoning can be pushed within a strictly small-model regime. Building upon the Spectrum-to-Signal post-training paradigm, we systematically enhance the model through an optimized pipeline that includes curriculum-based supervised fine-tuning, multi-domain reinforcement learning, and offline self-distillation. Experimental evaluations demonstrate that VibeThinker-3B achieves frontier-level performance on highly demanding verifiable tasks. Specifically, it attains a score of 94.3 on AIME26 (improving to 97.1 with claim-level test-time scaling), an 80.2 Pass@1 on LiveCodeBench v6, and exhibits strong out-of-distribution generalization with a 96.1\% acceptance rate on recent unseen LeetCode contests. This effectively places it in the performance band of first-tier reasoning systems, matching or exceeding flagship models that are orders of magnitude larger, such as DeepSeek V3.2, GLM-5, and Gemini 3 Pro. Furthermore, a score of 93.4 on IFEval confirms that this extreme reasoning enhancement does not compromise strict instruction controllability. Extending our previous 1.5B work, these findings motivate the Parametric Compression-Coverage Hypothesis, which views verifiable reasoning as compressible into compact reasoning cores, while open-domain knowledge and general-purpose competence require broad parameter coverage over facts, concepts, and long-tail scenarios. This perspective suggests that compact models are not merely deployment-efficient substitutes, but a complementary path toward frontier-level performance in parameter-dense capability regimes.
Who Is Affected
- DeepSeek
- Google DeepMind
- AI governance teams
- AI product teams
What To Watch Next
- Watch for regulator follow-through, court filings, compliance deadlines, and company policy changes.
- Watch whether additional sources confirm the same claim.
You will be redirected to huggingface.co.