Qwen / image-text-to-text updated 1 year ago

Qwen2.5-VL-3B-Instruct

Name: Qwen2.5-VL-3B-Instruct
Author: Qwen

--- licensename: qwen-research licenselink: https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct/blob/main/LICENSE language: - en pipelinetag: image-text-to-text tags: - multimodal libraryname: transformers ---

Params

3.8 B

Context

128,000

Downloads 30d

8.2 M

Likes

683

Licence: unknown Not gated SAFETENSORS 1 languages View on Hugging Face ↗

Download history

daily snapshots · 10 days

8.2 M7.7 M

Jul 28Jul 31Aug 3Aug 6

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

File	Quant	Size	Est. VRAM	Verdict on RTX 4090 · 24 GB
model.safetensors	bf16	7.5 GB	9.3 GB	✅ Runs comfortably
model.safetensors (bf16, full)	bf16 + 125K ctx	7.5 GB	17.5 GB	✅ Runs comfortably

Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Specifications

Architecture: Qwen2_5_VLForConditionalGeneration
Parameters: 3.8 B
Tensor type: BF16
Context length: 128,000
Vocabulary: 151,936
Layers / heads: 36 / 16
First seen on the Hub: 2025-01-26
Training datasets: undisclosed
Added to our catalog: 2026-07-28

Compare with

Qwen3.5-9B 12.4 M Qwen2.5-VL-7B-Instruct 9.3 M Qwen3.5-4B 6.7 M Qwen3-VL-4B-Instruct 4.0 M chandra-ocr-2 3.2 M Unlimited-OCR 2.7 M

Appears in

Best open coding models Commercial-use text models Models that fit in 12 GB