Skip to content
Local AI Wizard
Your machine
Theme

Step3-VL-10B

StepFun · stepfun-ai/Step3-VL-10B

Weights license: Apache 2.0 · code and components may use other terms

Reported by the publisher15 resultsStepFun's model card · read 2026-10-25 · not ranked · see them
Memory needed~8 GBQ4 weights + working room, before the machine's reserve
Parameters10.2 BHub safetensors total
Repository updated2026-02-04A commit is not a release date

Quality

No comparable score

We track no public, comparable benchmark that covers this model yet. No score is inferred.

Reported by the publisher

15 results · StepFun's model card · read 2026-10-25
MMMU
78.11STEM Reasoning
MathVision
70.81STEM Reasoning
MathVista
83.97STEM Reasoning
PhyX
59.45STEM Reasoning
MMBench (EN)
92.05Recognition
MMStar
77.48Recognition
ReMI
67.29Recognition
OCRBench
86.75OCR & Document
Every reported result15
AI2D
89.35OCR & Document
ScreenSpot-V2
92.61GUI Grounding
ScreenSpot-Pro
51.55GUI Grounding
OSWorld-G
59.02GUI Grounding
BLINK
66.79Spatial
All-Angles-Bench
57.21Spatial
HumanEval-V
66.05Code

Values as printed on the card, with the publisher's own settings (card section: 📊 Performance › Comparison with Open-Source Models (7B–10B)); column “STEP3-VL-10B”. Not comparable across publishers and not used in any ranking. Model card

Runs on

Fast memory holds it 90

The planning estimate fits the memory the model can use quickly: graphics card VRAM, or unified memory within the macOS GPU share.

Show all 90

Within memory, outside the fast part 26

Within the machine's memory, outside its fast part: on a graphics card the layers that do not fit run on the CPU from PC RAM (the listed RAM, or 32 GB assumed); on a Mac, above the macOS default GPU share. Not a speed figure.

Show all 26

One memory tier short 3

The same machines with the next memory size up would hold it in fast memory.

Hardware profiles grouped by the Q4 planning estimate on each machine (dedicated GPUs reserve 2 GB; unified and system memory 25%, 2 to 8 GB): the fast memory holds it, it is within memory but outside the fast part (layers offloaded to PC RAM, or above the macOS GPU share), or the machine is one memory tier short. Not a measured fit or speed.

Consider it for

StepFun released Step3-VL-10B on 2026-01-14: a vision-language model that answers about images, with 10.2B parameters on the Hugging Face count and the Apache 2.0 license.

Check first

License: commercial use allowed. At Q4 the weights load about 7.6 GB before each machine's reserve; image resolution, video frames and context add memory, and the vision encoder needs support in the runtime you pick.

Measured speed

No published measurement for this checkpoint

Tokens per second are shown only when a source publishes them with hardware, build and setup. They are never estimated.

How to run

Runtime compatibility and tokens per second have not been measured for this entry. Published metadata and a memory estimate do not prove a successful local run.

Documented ways to run it

Copied as printed in the linked source (read 2026-10-18); not run by this site.

Transformers · model card · Inference with Hugging Face Transformers
from transformers import AutoProcessor, AutoModelForCausalLM


key_mapping = {
    "^vision_model": "model.vision_model",
    r"^model(?!\.(language_model|vision_model))": "model.language_model",
    "vit_large_projector": "model.vit_large_projector",
}

model_path = "stepfun-ai/Step3-VL-10B"

processor = AutoProcessor.from_pretrained(model_path, trust_remote_code=True)

messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
            {"type": "text", "text": "What's in this picture?"}
        ]
    },
]

model = AutoModelForCausalLM.from_pretrained(
    model_path,
    trust_remote_code=True,
    device_map="auto",
    torch_dtype="auto",
    key_mapping=key_mapping).eval()


inputs = processor.apply_chat_template(
    messages, add_generation_prompt=True, tokenize=True,
    return_dict=True, return_tensors="pt"
).to(model.device)


generate_ids = model.generate(**inputs, max_new_tokens=1024, do_sample=False)
decoded = processor.decode(generate_ids[0, inputs["input_ids"].shape[-1] :], skip_special_tokens=True)

print(decoded)
vLLM · model card · 🚀 Deployment with vLLM (OpenAI-compatible API)
vllm serve --model stepfun-ai/Step3-VL-10B -tp 1 --reasoning-parser deepseek_r1 --enable-auto-tool-choice --tool-call-parser hermes --trust-remote-code
SGLang · model card · 🚀 Deployment with SGLang (OpenAI-compatible API)
sglang serve --model-path stepfun-ai/Step3-VL-10B --trust-remote-code --port 2345 --reasoning-parser deepseek-r1 --tool-call-parser hermes

Choose my computer

Technical details

What the source and the catalog store for this checkpoint. Missing fields are left out, never zero.

Checkpoint / source
stepfun-ai/Step3-VL-10B
Task
image-text-to-text
Architecture
StepVLForConditionalGeneration
Weight license
Apache 2.0stepfun-ai/Step3-VL-10B model card (Hugging Face) · read 2026-10-18source
Commercial use
Allowed Apache License 2.0 text (SPDX) · read 2026-10-18license text
Released
STEP3-VL-10B Technical Report (StepFun, arXiv 2601.09668 v1) · read 2026-10-18source
Every stored field12 fields
Revision checked
5026053b0c2f5dfaa08fc2d149384162c3c8bca1
Architecture
StepVLForConditionalGeneration
Hub task tag
image-text-to-text
Declared base models
stepfun-ai/Step3-VL-10B-Base
Formats
safetensors
Access
Open download
Publisher kind
Official
Hub parameter count
10,171,750,144safetensors.total
Repository created
2026-01-13creation is not the model's release
Repository updated
2026-02-04a commit is not a release date
Hub downloads (30 days) · likes
9,708 · 419this repository only · read 2026-10-25
Catalog state
Hub config or weights found; no inventory benchmark link

Sources

Identity and published specifications come from the checkpoint page. Benchmark scores belong to the named release and do not measure speed on your computer.

Curated facts checked: 2026-10-18. The license and release date cite the publisher's sources; parameters, files and task come from the Hugging Face record, and memory is a labelled estimate from its parameter count. A value no source states stays empty; a repository commit is not a model release date.