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

Skywork R1V

Skywork R1V is an open-source multimodal vision-language reasoning model series, with model weights and Python inference code for processing images alongside text questions.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Skywork R1V — 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 Skywork R1V when you want model weights and inference code to integrate image reasoning into a Python workflow, rather than a ready-made hosted assistant.

Who it is for

  • Developers building applications that answer questions about images.
  • Researchers and practitioners evaluating multimodal reasoning models.
  • Teams creating image-understanding workflows that can run model inference in their own environment.

Documented uses

  • Run a model on an image and a natural-language question.
  • Use the inference code and weights as the basis for a vision-language assistant.
  • Reproduce model evaluations using the repository's evaluation code.
Before you choose

Requirements & limits

The GitHub repository describes the code as MIT-licensed and permits commercial use, modification, and distribution.

  • The repository documents Python 3.10 environment setup and separate Transformers and vLLM inference paths.
  • The example commands use CUDA-visible GPUs; the vLLM example specifies tensor parallelism.
  • The repository provides model weights and inference code, so users must prepare an environment and supply model and image paths.
  • The README identifies InternVL3-38B as the base model and says it is MIT-licensed.

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

Prepare the local inference environment

Set up one of the repository's documented Python 3.10 environments before running inference.

Before you start

  • Clone access to the Skywork-R1V GitHub repository.
  • Conda and the repository's setup scripts.

Steps

  1. Clone the repository and change into its inference directory.

  2. For the Transformers path, create and activate a Python 3.10 Conda environment, then run bash setup.sh.

  3. For the vLLM/evaluation path, create and activate a Python 3.10 Conda environment, then run bash ./eval/vlmevalkit/build_env.sh.

Expected outcome

A prepared environment for the selected inference or evaluation path; the instructions do not themselves run a model.

Guide limitations

  • The documented setup stage does not include model execution.
  • The instructions do not specify a general hardware configuration for all model variants.
Documentation 1 ↗
Source-based practical guide

Run image-question inference

Submit an image and a question to a locally configured Skywork R1V model using the documented inference script.

Before you start

  • A prepared Python 3.10 environment for Transformers or vLLM.
  • A local model path and image path or paths.
  • CUDA-visible GPU resources for the example commands.

Steps

  1. Choose the Transformers or vLLM inference script.

  2. Set the model path, image path or paths, and question in the command arguments.

  3. For the Transformers example, set CUDA_VISIBLE_DEVICES; for vLLM, provide the tensor-parallel-size argument.

Expected outcome

The inference script is invoked with image input and a text question to produce a model response.

Guide limitations

  • The repository examples require adapting placeholder paths and question text.
  • The vLLM example uses a tensor-parallel size of 4, while the Transformers example exposes two GPU IDs; these examples do not establish universal resource requirements.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Model inference and serving

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
Skywork R1VRun a model on an image and a natural-language question.The GitHub repository describes the code as MIT-licensed and permits commercial use, modification, and distribution.
Pulse Of MotionEstimate a video's physical frame rate from its visual dynamics.LinuxThe repository is public. Its README describes a released checkpoint that downloads from Hugging Face on first inference, with a manual download option. The strongest internal model is not fully released; the README directs interested users to contact the project about access. No license information is provided in the repository evidence.
Neural Image AssessmentTrain the model using AVA images and rating annotations.Cloud, Linux, macOS, Windows
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
10 October 2026
Original website title
SkyworkAI/Skywork-R1V · Official homepage title · retained in its source language
How we review sourcesSuggest a correction