VibeThinker-3B
WeiboAIWeiboAI's VibeThinker-3B, an MIT-licensed 3B reasoning model fine-tuned from Qwen2.5-3B / Qwen2.5-Coder-3B and specialized for verifiable math, coding and STEM. Claims frontier-range scores on narrow reasoning benchmarks despite its size; not intended for tool-calling or agentic use.
MIT-licensed weights, but a fine-tune of the open-weight Qwen2.5 line without a fully open data/training pipeline -> open_weights, not open_source. Narrow specialist. The adoption axis records reported_traction while a countable Hub artifact exists; flagged for re-banding. Verified 2026-08-13 via the HF model card.
Openness
3 medium confidence- weights
- open(MIT)
- base
- open_weights(Qwen2.5)
- data
- partial(RL/SFT data not fully released)
- license
- MIT
MIT-licensed weights, but a fine-tune of the open-weight Qwen2.5 line without a fully open or reproducible data and training pipeline, so open weights rather than open source.
- https://huggingface.co/WeiboAI/VibeThinker-3B recorded 2026-08-13
`license:mit` in the card data and `"gated":false`; `base_model - Qwen/Qwen2.5-Coder-3B`; no RL/SFT mixture and no training pipeline published
Adoption
3 high confidenceA recently released specialist model. Its Hugging Face model repository returns 149,008 downloads in the trailing 30 days, which lands in 100K-1M on the model scale. There is a real count behind the band rather than a market-position judgment, which is why confidence is high.
- https://huggingface.co/api/models/WeiboAI/VibeThinker-3B recorded 2026-08-14
149,008 downloads in the trailing 30 days for WeiboAI/VibeThinker-3B, ungated
- https://huggingface.co/WeiboAI/VibeThinker-3B recorded 2026-08-13
active HF repo; 822 likes; early traction as a small reasoning specialist
Capability
3 medium confidenceFrontier-range on narrow, verifiable math and code reasoning despite being only 3B, but explicitly not built for general or agentic use, so mid-tier on a general-capability scale.
- https://huggingface.co/WeiboAI/VibeThinker-3B recorded 2026-08-13
model-index - MathArena/aime_2026 94.3, Idavidrein/gpqa diamond 70.2; Key Performance Data - "VibeThinker-3B reaches 76.4 on IMO-AnswerBench ... with only 3B parameters"; not positioned for tool or agent use
Verified 2026-08-13