CASA-Qwen2_5-VL-3B

kyutai/CASA-Qwen2_5-VL-3B

Image Text to Text4.1B parameterstransformersLicense: cc-by-nc-sa-4.0

Tracked by OpenModelStats since September 6, 2026View on Hugging Face ↗

Downloads 30D: 102
Total downloads: 1,005
Gained 7D: 0; 0% vs baseline
Gained 30D: —
Likes: 3; +0 in 7D
Rank · tracked models: —; –this week
Parameters: 4.1B
Last source update: Sep 6, 2026

7D 30D 90D 1Y All

Total downloads Gained per day Rolling recent downloads

Downloads over time

Rolling recent-window downloads as reported

Likes over time

Rank among tracked models

Lower is better

Not enough tracked history yet for a chart.

OpenModelStats began tracking this model on 2026-09-06. Charts appear as daily observations accumulate.

Overview

CASA-Qwen2_5-VL-3B is a image text to text model published by kyutai. OpenModelStats has tracked the model since Sep 6, 2026, most recently observing it Sep 7, 2026.

Published: Dec 10, 2025
Last updated: Sep 6, 2026
Library: transformers
License: cc-by-nc-sa-4.0
Task: Image Text to Text
Parameters: 4,093,774,848
Spaces: —
Derivative models: —
Velocity 7D/day: 0

transformerssafetensorsqwen2_5vl_cafeature-extractionimage-text-to-textconversationalcustom_codeendataset:HuggingFaceM4/FineVisiondataset:mvp-lab/LLaVA-OneVision-1.5-Instruct-Dataarxiv:2512.19535base_model:Qwen/Qwen2.5-VL-3B-Instruct

Publisher

kyutaiView publisher statistics →

Derived from Qwen/Qwen2.5-VL-3B-Instruct (reported by the source; parent not yet tracked).

Similar models

Models similar to CASA-Qwen2_5-VL-3B
Model	Task	DL 30D	Likes	Params
POINTS-Readertencent	Image Text to Text	312	103	4.1B
Qwen2.5-VL-3B-Instruct-quantized.w8a8RedHatAI	Image Text to Text	25.4K	2	4.1B
Qwen2.5-VL-3B-Instruct-FP8-dynamicRedHatAI	Image Text to Text	8,164	3	4.1B
Dharma-OCR-LITEDharma-AI	Image Text to Text	586	22	4.1B
Rex-OmniIDEA-Research	Image Text to Text	12.2K	64	4.1B
ViGoRL-3b-Spatialgsarch	Image Text to Text	1,341	2	4.1B