Skip to content

Alibaba · Chat

Qwen2.5 VL 72B Instruct

Alibaba 🇨🇳ChatQwen familyOpen weights. One of its sources says the weights are publicly available.The maker lists this model as deprecated (5 Jan 2026).
From Cortecs AI's listing:
Qwen2.5-VL is a powerful vision-language model with advanced capabilities in visual understanding, long video reasoning, and structured output generation.
llminfo IQ
108.8
Market median price
$0.85 blended / 1M
Max context
128K
Listings
26

Evidence and sources

International storefront (alibabacloud.com) is Singapore-incorporated; mainland operating entity 杭州阿里云智能科技有限公司 (Hangzhou).

🤗 Qwen/Qwen2.5-VL-72B-Instructllminfo IQ 108.8, ranked among chat models. From 97 benchmarks. Plenty of data behind it.How llminfo IQ is computedPoor value (41.2). Costs more per task than its benchmark results predict. From 2 benchmarks with measured task costs. Little data behind it, so treat it as a rough estimate.How Value for money worksReleased 27 Jan 2025 Training-data cutoff, as its sources state it Maker HQ: CN (source)
All rankings: #26 in Hacker News discussions this week #56 on NanoGPT recently #293 on OpenRouter today #313 on OpenRouter this week #320 on OpenRouter this monthAll rankings: #26 in Hacker News discussions this week #56 on NanoGPT recently #293 on OpenRouter today #313 on OpenRouter this week #320 on OpenRouter this monthSources: LMArena data, OVH Cloud data, LiteLLM data, Models.dev data, OpenRouter data docs, Cortecs EU Router data, Hugging FacePrimary data from LMArena from OVH Cloud: name from LiteLLM: deprecation date from Models.dev: weights availability from OpenRouter: knowledge cutoff from Cortecs EU Router: description from Hugging Face: release date
Tags from its sources:instructvisionopen-sourcemultimodaldocument-analysis

Advantages and trade-offs

Compared with 10 chat models, picked mainly by popularity and IQ

Advantages

Open weights available

The weights are public. 2 of the 7 peers that say either way have open weights.

Trade-offs

Deprecated by its maker

The maker has already deprecated this model (5 Jan 2026).

We only make a claim when there's enough data on models of the same type. Updated 22 Nov, 06:07 UTC.

Input & output

as stated by the model's providers
Text
Images
Audio
Video
Text

Capabilities

Supported operations and provider-stated features

VisionFunction callingStructured outputPrompt caching

Listings

 OpenRouterqwen/qwen2.5-vl-72b-instruct:free · freeSource: Helicone Costs dataNot published by the provider$0.00$0.00
 Flowdotqwen/qwen2.5-vl-72b-instructSource: FlowDot data32K$0.25$0.75
 Hugging FaceInference by Nebius AI StudioQwen/Qwen2.5-VL-72B-InstructSource: GenAI Prices data32K$0.25$0.75
 Nebius AI Studionebius/Qwen/Qwen2.5-VL-72B-InstructSource: LiteLLM data32K$0.25$0.75
 AihubmixQwen/Qwen2.5-VL-72B-InstructSource: AIHubMix data docsNot published by the provider$0.50$0.50
 302aiqwen/qwen2.5-vl-72b-instructSource: 302.ai data128K$0.60$0.60
 NanoGPTqwen/qwen2.5-vl-72b-instructSources: Models.dev data, NanoGPT32K$0.6999cache read $0.35$0.6999
 RequestyInference by Parasailparasail/parasail-qwen25-vl-72b-instructSource: Requesty data33K$0.70$0.70
 Kiloqwen/qwen2.5-vl-72b-instructSources: Kilo AI Gateway data, Models.dev data128K$0.80cache read $0.40$1.00
 Wavespeedqwen/qwen2.5-vl-72b-instructSource: WaveSpeed AI data128K$0.80cache read $0.40$1.00
Show all 23 listings

Showing 10 of 23: the cheapest per 1M input tokens from each provider.

Also sold by Cortecs AI, Eurouter, Fireworks AI, Novita AI, OVHcloud, Qiniu, Together AI, Tokenmix.

3 listings without a price are hidden.

Benchmarks

Named effort levels run from lowest to highest. Select a result to see its source details.

Vision

General

Spatialviz BenchSource: Epoch AI Benchmarking Hub data

Reasoning

Human preference

Science

GPQA Diamond3 results
Source: OpenRouter benchmark exports · Sep 2026
qwen2.5-vl-72b-instruct · 90-day results
Across 2 providers lowest 44.3% median 45% highest 45.7%OpenRouter's automatic routing 42.9%
Show results from 2 providers and OpenRouter's automatic routingHide results from 2 providers and OpenRouter's automatic routing
  • Parasail
  • Nebius
  • OpenRouter's automatic routing

Agentic

Knowledge

Other

Data credits: LMArena, CC BY 4.0 (licence) · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · OpenRouter public benchmark leaderboard and search-lane exports, CC BY 4.0. Fractional source scores converted to percentages; immutable dated snapshots. (licence) · Hugging Face official benchmark leaderboards; licence varies by benchmark · Steel.dev Agent Leaderboards, MIT (licence) · Physical AI Bench (SHI Labs), MIT (licence) · ScreenSpot grounding leaderboard (GUI-Agent), Apache-2.0 (licence) · SnakeBench (Greg Kamradt / ARC Prize Foundation), MIT (licence)

Around the web

Also known as: fireworks_ai/accounts/fireworks/models/qwen2p5-vl-72b-instruct, nebius/qwen/qwen2.5-vl-72b-instruct, novita/qwen/qwen2.5-vl-72b-instruct, ovhcloud/qwen2.5-vl-72b-instruct, parasail/parasail-qwen25-vl-72b-instruct, qwen-2.5-vl-72b-instruct, qwen/qwen-2.5-vl-72b-instruct, qwen/qwen2.5-vl-72b, qwen/qwen2.5-vl-72b-instruct, qwen/qwen2.5-vl-72b-instruct:free, qwen2.5-vl-72b, qwen25-vl-72b-instruct, qwen2p5-vl-72b-instruct, sophnet/qwen2.5-vl-72b-instruct, together_ai/qwen/qwen2.5-vl-72b-instruct

Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works