Qwen3-VL-235B-A22B

Multimodal

by Alibaba / Qwen

+
+
+
+
About

Qwen3-VL-235B-A22B, released by Alibaba's Qwen team in September 2025, is a natively multimodal MoE model from the Qwen3 visual language series — designed for advanced image and video understanding tasks. It extends the Qwen3 architecture with native visual processing, targeting applications that require both strong language capabilities and visual reasoning at the scale of the Qwen3-235B family.

+
+
+
+
Pricing Range
Input (per 1M)$0.25 -$0.25
Output (per 1M)$0.75 -$0.75
Providers1
+
+
+
+
Timeline
ReleasedSep 23, 2025
+
+
+
+
Specifications
Capabilities
Multimodal
+
+
+
+
License & Family
License
Apache 2.0
Performance Overview
Performance metrics and category breakdown

Overall Performance

1 benchmarks
Average Score
68.1%
Best Score
68.1%
High Performers (80%+)
0

Performance Metrics

Max Context Window
270.3K

Top Categories

Multimodal
68.1%
+
+
+
+
All Benchmark Results for Qwen3-VL-235B-A22B
Complete list of benchmark scores with detailed information
MMMU
Multimodal
68.10
68.1%
Unverified