VibeThinker 3B — VRAM, speed & local setup
Written by Jakub Rusinowski · Last updated
A 3B dense reasoner built on Qwen2.5-Coder-3B. Reports 94.3 on AIME26, 89.3 on HMMT25 and 76.4 on IMO-AnswerBench — figures that put it in frontier-reasoner territory on competition maths while staying under 3 GB at Q4_K_M. Those numbers are the authors' own and remain disputed; the model is worth running precisely because you can check them yourself on a laptop.
VibeThinker 3B needs about 3 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
VRAM and speed by quantization
Quoted against NVIDIA RTX 4090 (24 GB). Weights plus overhead only: this model's architecture is unpublished, so no KV cache is added.
| Quant | Memory | VRAM | Speed (est.) | Fit |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 1.8 GB | ~241 tok/s | Fits | |
| Q3_K_M 3.41 bpw | 2.1 GB | ~220 tok/s | Fits | |
| Q4_K_M 4.83 bpw | 2.7 GB | ~190 tok/s | Fits | |
| Q5_K_M 5.67 bpw | 3 GB | ~176 tok/s | Fits | |
| Q6_K 6.56 bpw | 3.3 GB | ~163 tok/s | Fits | |
| Q8_0 8.50 bpw | 4.1 GB | ~140 tok/s | Fits | |
| F16 16.00 bpw | 7 GB | ~91 tok/s | Fits |
Black marker = usable memory on the NVIDIA RTX 4090 (24 GB). Estimates from the memory-bandwidth roofline on the methodology page. VibeThinker 3B VRAM calculator →
Get VibeThinker 3B running
The cheapest catalogued GPU that runs VibeThinker 3B is the Intel Arc B570 (10 GB).
or compare on Vast.ai from $0.35/hr (typical low · varies)
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
How to run VibeThinker 3B
Install Ollama, then run:
ollama run vibethinkerSpecifications
Verified — Checked against the primary source — the model card or the vendor spec page — and corroborated by a second independent source.
- Parameters
- 3.09 Billion
- Context window
- 131,072
- Architecture
- Dense Transformer (Qwen2.5-Coder-3B base)
- Provider
- WeiboAI
- Licence
- MIT
- Practical deployment
- Edge device (phone, SBC)
- Specified at
- Q4_K_M
- System RAM
- 8 GB
- Record updated
- 2026-09-06
Commercial use permitted. No usage restrictions beyond attribution.
Quality and use cases
Scores as published by the model’s authors or an independent evaluator — quality, not throughput, and not measured by us.
| Benchmark | Score | Provenance |
|---|---|---|
| AIME 2026 | 94.3 / 100 % | vendor-claimed · https://huggingface.co/WeiboAI/VibeThinker-3B |
| HMMT 2025 | 89.3 / 100 % | vendor-claimed · https://huggingface.co/WeiboAI/VibeThinker-3B |
| IMO-AnswerBench | 76.4 / 100 % | vendor-claimed · https://huggingface.co/WeiboAI/VibeThinker-3B |
Can I run VibeThinker 3B on my GPU?
Other VibeThinker sizes
VibeThinker 3B — frequently asked questions
How much VRAM does VibeThinker 3B need?
About 3 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does VibeThinker 3B run on an RTX 4090 (24 GB)?
Yes. VibeThinker 3B needs about 3 GB at Q4_K_M, inside a 24 GB card, at an estimated 190 tokens/sec.
How do I run VibeThinker 3B locally?
Install Ollama and run `ollama run vibethinker`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.
What other sizes does VibeThinker come in?
VibeThinker 3B (3 GB), VibeThinker 1.5B (2 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.