inclusionAI 发布 VISTA-4B GUI 定位视觉语言模型

VISTA-4B are GUI-grounding vision-language models trained from Qwen3.5 4B backbones with VISTA: View-Consistent Self-Verified Training for GUI Grounding.

Model Description

VISTA-4B is a GUI-grounding model that maps a screenshot and a natural-language instruction to a click coordinate in the normalized 0-1000 image frame.

View-consistent GRPO training. VISTA builds each GRPO comparison group from target-preserving views of the same GUI instance, with exact coordinate remapping across cropped views. This exposes localization behavior under semantically equivalent but geometrically different screenshots.
Self-verified cross-view anchoring. The training objective adds an oracle-format center-point anchor only when model-generated rollouts have already produced a maximum-reward prediction, stabilizing short coordinate generation without unconditional imitation on all-fail groups.

Evaluation

Accuracy is reported for GUI grounding. The model predicts a normalized coordinate in the 0-1000 frame, and the prediction is counted as correct if the point lies inside the target element. All reported results use deterministic decoding at temperature 0 and single-view inference.

Results on GUI Grounding benchmarks

Model	SSPro	SSV2	OSWorld-G	OSWorld-G-R
Qwen3.5-4B	60.3	90.4	54.4	66.8
GRPO-4B	62.2	94.2	59.9	69.2
VISTA-4B	64.2	93.8	61.2	69.7
Δ	+2.0	-0.4	+1.3	+0.5
Qwen3.5-9B	65.2	91.9	63.1	74.6
GRPO-9B	68.3	95.2	67.5	75.2
VISTA-9B	69.2	95.8	68.1	75.5
Δ	+0.9	+0.6	+0.6	+0.3
Qwen3.5-35B-A3B	68.6	93.8	65.8	72.5
GRPO-35B-A3B	71.7	95.7	70.4	74.3
VISTA-35B-A3B	72.9	95.8	71.5	75.3
Δ	+1.2	+0.1	+1.1	+1.0

Quick Start

Use the same image-chat interface as the underlying Qwen3.5 vision-language model. The recommended prompt is:

Output the center point of the position corresponding to the instruction: {instruction}. The output should just be the coordinates of a point, in the format [x,y].

Example:

import torch
from PIL import Image
from transformers import AutoModelForImageTextToText, AutoProcessor

model_id = "inclusionAI/VISTA-4B"  

model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
)
processor = AutoProcessor.from_pretrained(model_id, trust_remote_code=True)

image = Image.open("screenshot.png").convert("RGB")
instruction = "Click the search button"
prompt = (
    "Output the center point of the position corresponding to the instruction: "
    f"{instruction}. The output should just be the coordinates of a point, "
    "in the format [x,y]."
)

messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "image": image},
            {"type": "text", "text": prompt},
        ],
    }
]

text = processor.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)
inputs = processor(
    text=[text],
    images=[image],
    padding=True,
    return_tensors="pt",
).to(model.device)

generated = model.generate(
    **inputs,
    max_new_tokens=32,
    do_sample=False,
)
new_tokens = generated[:, inputs.input_ids.shape[1]:]
response = processor.batch_decode(new_tokens, skip_special_tokens=True)[0].strip()
print(response)  # e.g. [512,384]

Citation

Please consider citing if you find our work useful:

@misc{qiu2026vista,
      title={VISTA: View-Consistent Self-Verified Training for GUI Grounding},
      author={Xinyu Qiu, Yunzhu Zhang, Heng Jia, Shuheng Shen, Changhua Meng, Linchao Zhu},
      year={2026},
      eprint={2606.14579},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2606.14579},
}