README.md

metadata

license: mit
library_name: transformers
pipeline_tag: image-text-to-text

Skywork-R1V3-38B-AWQ

📖 R1V3 Report | 💻 GitHub | 🌐 ModelScope

Evaluation

Comprehensive performance comparison across text and multimodal reasoning benchmarks.

Model	MMMU	MathVista
Proprietary Models
Claude-3.7-Sonnet	75.0	66.8
OpenAI-4o	70.7	62.9
Open-Source Models
InternVL3-78B	72.2	72.2
Qwen2.5-VL-72B	70.3	74.8
QvQ-Preview-72B	70.3	71.4
Skywork-R1V3	76.0	77.1
Skywork-R1V3-AWQ	66.7	70.5

Usage

You can use the quantized model with different inference frameworks:

Using VLLM

Python API

import os
from vllm import LLM, SamplingParams
from vllm.entrypoints.chat_utils import load_chat_template
model_name = "Skywork/Skywork-R1V3-38B-AWQ"  # or local path
llm = LLM(model_name, 
          dtype='float16', 
          quantization="awq", 
          gpu_memory_utilization=0.9,
          max_model_len=4096,
          trust_remote_code=True,
         )
# Add your inference code here

OpenAI-compatible API Server

MODEL_ID="Skywork/Skywork-R1V3-38B-AWQ"  # or local path
CUDA_VISIBLE_DEVICES=0 \
    python -m vllm.entrypoints.openai.api_server \
    --model $MODEL_ID \
    --dtype float16 \
    --quantization awq \
    --port 23334 \
    --max-model-len 12000 \
    --gpu-memory-utilization 0.9 \
    --trust-remote-code

Using LMDeploy

import os
from lmdeploy import pipeline, TurbomindEngineConfig, ChatTemplateConfig
from lmdeploy.vl import load_image
model_path = "Skywork/Skywork-R1V3-38B-AWQ"  # or local path
engine_config = TurbomindEngineConfig(cache_max_entry_count=0.75) 
chat_template_config = ChatTemplateConfig(model_name=model_path)
pipe = pipeline(model_path, 
                backend_config=engine_config, 
                chat_template_config=chat_template_config,
               )
# Example: Multimodal inference
image = load_image('table.jpg')
response = pipe(('Describe this image?', image))
print(response.text)

Hardware Requirements

The AWQ quantization reduces the memory footprint compared to the original FP16 model. We recommend:

• At least one GPU with 30GB+ VRAM for inference
• For optimal performance with longer contexts, 40GB+ VRAM is recommended

Citation

If you use this model in your research, please cite:

@misc{shen2025skyworkr1v3technicalreport,
      title={Skywork-R1V3 Technical Report}, 
      author={Wei Shen and Jiangbo Pei and Yi Peng and Xuchen Song and Yang Liu and Jian Peng and Haofeng Sun and Yunzhuo Hao and Peiyu Wang and Jianhao Zhang and Yahui Zhou},
      year={2025},
      eprint={2507.06167},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2507.06167}, 
}