Qwen-VL Max

Qwen-VL-Max is a proprietary vision-language model from Alibaba's QwenLM team, positioned as the top tier in the Qwen-VL family above the VL-Plus variant. It is designed for sophisticated multimodal applications including document parsing, visual reasoning, multilingual analysis, and structured data extraction. Unlike the open-weight Qwen2.5-VL series that includes models scaling up to 72B parameters, Qwen-VL-Max itself is not available as open weights, reflecting its role as a premium proprietary offering within the broader Qwen ecosystem.

In practical deployment through Alibaba Cloud Model Studio, Qwen-VL-Max is being upgraded to a stable snapshot dated August 2025, with the Beijing and Singapore regions both receiving the updated version while maintaining existing pricing structures. The Singapore deployment adds batch inference support priced at half the standard input rate. With its large context capacity and ability to handle text and image inputs, Qwen-VL-Max fits workflows requiring deep visual understanding of complex documents and images, making it suitable for enterprises needing high-fidelity multimodal analysis without managing open-weight infrastructure themselves.

Ofoxqwen/qwen-vl-maxqwen

Open provider page Provider docs

Quick Info

Provider: Ofox
Model key: qwen/qwen-vl-max
Release date: Apr 8, 2024
Last updated: Aug 13, 2025
Knowledge cutoff: 2024-04
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.23
Output token cost: $0.58

Limits

Output tokens: 8,000 tokens
Context window: 128,000 tokens

Transparent token rates

Compare Qwen-VL Max pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about Qwen-VL Max

Eden AI

CoverageBenchmark

BenchmarkList's dedicated Qwen VL Max page surfaces the model's ranking on the VeriTrip agentic benchmark (Agentic1 eval, 30th percentile), where it places 8th out of 11 with a DR (Simple) score of 61.9% and an FR (Simple) score of 42.1%. The page positions these results alongside competing models including Qwen O-5, C The same BenchmarkList entry lists per-1M-token pricing figures for Qwen-VL Max alongside its peers, offering concrete cost-reference data for developers. Because the pricing figures appear to reflect aggregator/gateway-tier rates rather than Alibaba's own list price, they should be treated as observational aggregator

Eden AI

Coverage

Roboflow's Playground model page explicitly profiles Qwen-VL Max, confirming it as a proprietary vision-language model from Alibaba's QwenLM team released on February 1, 2025, and positioned as the flagship offering in the Qwen-VL family above the VL-Plus tier. The page documents that Qwen-VL Max accepts text and image The same source lists the model as closed/proprietary with multimodal modality and surfaces Roboflow Playground usage telemetry of 49 inferences over the prior 30 days at an average latency of 7.48 seconds. It highlights target use cases including document parsing, visual reasoning, multilingual analysis, and structure