OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Vibethinker

Vibethinker is the public repository for the VibeThinker-1.5B and VibeThinker-3B reasoning language models, with links to their weights, technical reports and usage guidance.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Vibethinker — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Vibethinker when you want to inspect or work with its compact reasoning models for competitive-style mathematics, coding, STEM reasoning or constrained instructions, and can provide an inference environment.

Who it is for

  • Developers integrating a reasoning model into an application with a separately selected inference setup.
  • Researchers and evaluators studying compact models on mathematics, coding, STEM and other verifiable tasks.

Documented uses

  • Review VibeThinker model reports and papers.
  • Access model weights linked by the repository.
  • Use the documented Transformers example to generate text from a model path.
  • Evaluate model outputs on tasks where answers can be verified.
Before you choose

Requirements & limits

  • The repository offers two model sizes, 1.5B and 3B parameters.
  • Transformers 4.54.0 or later is required for the documented quick-start approach.
  • The project recommends vLLM 0.10.1 or SGLang 0.4.9.post6 or later for better inference performance.
  • The repository says larger general-purpose models may be more suitable for broad open-domain knowledge tasks.
  • The repository metadata lists an MIT license.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Load a VibeThinker model with Transformers

Use the repository's documented Transformers example to generate a response from a local model path.

Before you start

  • Transformers 4.54.0 or later
  • A VibeThinker model path
  • A Python environment with the dependencies required by the example

Steps

  1. Obtain a VibeThinker model path from the model weights linked by the repository.

  2. Load the model with AutoModelForCausalLM.from_pretrained and its tokenizer with AutoTokenizer.from_pretrained, following the README example.

  3. Format a user prompt with the tokenizer's chat template and generate a response using the example's generation configuration.

Expected outcome

The example returns decoded model-generated text for the supplied prompt.

Guide limitations

  • The README does not specify hardware requirements.
  • vLLM 0.10.1 or SGLang 0.4.9.post6 or later are recommended for better inference performance, not required by the shown Transformers example.
  • The example uses trust_remote_code=True when loading the tokenizer.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
11 October 2026
Original website title
WeiboAI/VibeThinker · Official homepage title · retained in its source language
How we review sourcesSuggest a correction