Language modelActive

Qwen 2.5 VL 72B

Alibaba

Released: January 27, 2025
Data date: September 8, 2026
Successor: Qwen 3 VL 32B

Price: $0.8 · $1
Input · Output / 1M tokens
Input: Text, Image, Video
Output: Text

Qwen 2.5 VL 72B does more than describe a single image. It analyzes documents, diagrams, tables, user interfaces, and video, recognizes text across orientations, and can localize relevant visual regions. Bounding boxes and points can be returned as structured JSON.

The 72B model accepts text, images, and video and returns text. Qwen also describes long-video understanding and the ability to locate relevant events in time. In the technical report, Qwen 3 VL 32B is the medium-sized model explicitly compared with this checkpoint and reported to surpass it on visual reasoning tasks.

Specifications and access

Specification	Value and source
Weights	Open weightsSource
Access	API, LocalSource
Parameters	72BSource
Context window	128,000 tokensSource
Weights repository	Qwen/Qwen2.5-VL-72B-InstructSource
License	qwen (The model card declares a non-SPDX custom/other license; inspect the linked LICENSE before publishing a generic license badge.)Source
Architecture	ViT vision encoder + MLP vision-language merger + Qwen language modelSource
Input	Text, Image, VideoSource
Output	TextSource
Reasoning mode	instructSource
Tool calling	YesSource
Structured output	YesSource
Native context window	32,768 tokensSource
Maximum context window	32,768 tokensSource
Parameters	72 BSource
Languages	multilingual; exact count not stated in cardSource
Training tokens	4.1 trillionSource
Repository created	2025-01-27 (Repository creation date is a precise publication proxy, but may differ from the public announcement date or an individual checkpoint's first availability.)Source

Measurements without matching peer values

These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.

MMMU (Show measurement, test conditions, and source)

Source value: 70.2
Score: 70.2
Metric: percent
Unit: %
Category: visual
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen2.5-VL Technical Report
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official report/card benchmark table; exact model column

Source Qwen2.5-VL Technical Report

MMMU-Pro (Show measurement, test conditions, and source)

Source value: 51.1
Score: 51.1
Metric: percent
Unit: %
Category: visual
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen2.5-VL Technical Report
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official report/card benchmark table; exact model column

Source Qwen2.5-VL Technical Report

MathVista mini (Show measurement, test conditions, and source)

Source value: 74.8
Score: 74.8
Metric: percent
Unit: %
Category: visual
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen2.5-VL Technical Report
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official report/card benchmark table; exact model column

Source Qwen2.5-VL Technical Report

MathVisionFULL (Show measurement, test conditions, and source)

Source value: 38.1
Score: 38.1
Metric: percent
Unit: %
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen official Hugging Face model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official model-card table; shots/tools/decoding not stated in the card row

Source Qwen official Hugging Face model card

Hallusion Bench (Show measurement, test conditions, and source)

Source value: 55.16
Score: 55.16
Metric: percent
Unit: %
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen official Hugging Face model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official model-card table; shots/tools/decoding not stated in the card row

Source Qwen official Hugging Face model card

MMBenchDEVENV11 (Show measurement, test conditions, and source)

Source value: 88
Score: 88
Metric: percent
Unit: %
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen official Hugging Face model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official model-card table; shots/tools/decoding not stated in the card row

Source Qwen official Hugging Face model card

AI2DTEST (Show measurement, test conditions, and source)

Source value: 88.4
Score: 88.4
Metric: percent
Unit: %
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen official Hugging Face model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official model-card table; shots/tools/decoding not stated in the card row

Source Qwen official Hugging Face model card

ChartQATEST (Show measurement, test conditions, and source)

Source value: 89.5
Score: 89.5
Metric: percent
Unit: %
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen official Hugging Face model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official model-card table; shots/tools/decoding not stated in the card row

Source Qwen official Hugging Face model card

DocVQAVAL (Show measurement, test conditions, and source)

Source value: 96.4
Score: 96.4
Metric: percent
Unit: %
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen official Hugging Face model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official model-card table; shots/tools/decoding not stated in the card row

Source Qwen official Hugging Face model card

MMStar (Show measurement, test conditions, and source)

Source value: 70.8
Score: 70.8
Metric: percent
Unit: %
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen official Hugging Face model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official model-card table; shots/tools/decoding not stated in the card row

Source Qwen official Hugging Face model card

MMVetturbo (Show measurement, test conditions, and source)

Source value: 76.19
Score: 76.19
Metric: percent
Unit: %
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen official Hugging Face model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official model-card table; shots/tools/decoding not stated in the card row

Source Qwen official Hugging Face model card

CC-OCR (Show measurement, test conditions, and source)

Source value: 79.8
Score: 79.8
Metric: percent
Unit: %
Direction: Higher is better
Source type: vendor-reported
Evaluator: Qwen official Hugging Face model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Official model-card table; shots/tools/decoding not stated in the card row

Source Qwen official Hugging Face model card

Page 1 of 2

Predecessor and successor

Later model generation: Qwen 3 VL 32B
Next generation in the directory
Model lineage evidence
Section 5.2 (Multimodal Reasoning) names Qwen3-VL-32B as the subject of the paragraph ('Among medium-sized models... Qwen3-VL-32B demonstrates significant advantages') and then states 'Compared to the previous-generation Qwen2.5-VL-72B model, the medium-sized Qwen3-VL model has already surpassed it on reasoning tasks,' the only 'previous-generation' statement in the full paper text and the only Qwen2.5-VL-72B comparison tied to one specific Qwen3-VL size rather than the series generally; the sibling Qwen3-VL-30B-A3B MoE model is named separately in the same paragraph with no predecessor claim.
Qwen3-VL Technical Report (2026-09-19)

Sources and data date

Type	Evidence and data date
Price	Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. Source: Pricing · API pricing
Input	Documented modalities. Retrieved September 8, 2026. Source: Model documentation · Model metadata (retrieved September 8, 2026) · Qwen2.5-VL-72B-Instruct model card · Editorial description reviewed October 2, 2026
Output	Documented modalities. Retrieved September 8, 2026. Source: Model documentation · Model metadata (retrieved September 8, 2026) · Qwen2.5-VL-72B-Instruct model card · Editorial description reviewed October 2, 2026
Research status	Research date September 8, 2026. 4 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Not found in the inspected sources	active-parameters-billions Original wording Dense checkpoint; no separate activated-parameter count is applicable/published.
Unresolved	max-context-tokens Original wording The checked card/report configure 32,768 tokens and document YaRN extension, but do not state an exact tested maximum; the inventory's 128K value was not used as primary evidence.
Unresolved	AA benchmark rows Original wording No exact Artificial Analysis model page located in the checked public sitemap.
Additional source	Qwen2.5-VL Technical Report (retrieved September 8, 2026)
Additional source	Model metadata (retrieved September 8, 2026)
Additional source	Qwen2.5-VL official release · Editorial description reviewed October 2, 2026
Additional source	OpenRouter: Popularity index (derived), CC BY 4.0
Additional source	Vercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional source	Creative Commons Attribution 4.0
Additional source	Qwen3-VL Technical Report (2026-09-19)

Qwen 2.5 VL 72B

Specifications and access

Measurements without matching peer values

Our tests

Pelican on a bicycle

Popularity

Published prices

Predecessor and successor

Sources and data date