Qwen3 VL 235B A22B Instruct
Alibaba · Multimodal · Released Sep 2025
A 235-billion-parameter multimodal vision-language model from Alibaba that processes text, images, and video with a 131k-token context window.
- Strengths
- Handles large-scale multimodal tasks with its 235B parameter count, enabling sophisticated reasoning across text, images, and video in a single forward pass.
- Best for
- Complex vision-language applications requiring dense model capacity, such as detailed image analysis, multi-page document understanding, and video comprehension at scale.
- Limitations
- The 131k-token context window is shorter than some dense alternatives like Qwen3 Next 80B A3B Instruct, and extended reasoning workloads may benefit from the Thinking variant released concurrently.
Input / 1M
$0.21
Output / 1M
$1.90
Cached input / 1M
$0.1
Context window
131K
Price history
Input (solid)Output (dashed)
Price change
- 30d
- in — out —
- 90d
- in decreased 0.0% out decreased 0.0%
- 1y
- in increased 5.0% out increased 115.9%
- Since launch
- in increased 5.0% out increased 115.9%
Snapshots
| Effective | Input | Output | Cached in | Note | Source |
|---|---|---|---|---|---|
| 17 Aug 2026 | $0.21 | $1.90 | $0.1 | Imported from OpenRouter | openrouter.ai |
| 12 Aug 2026 | $0.26 | $1.04 | — | Imported from OpenRouter | openrouter.ai |
| 15 Jul 2026 | $0.21 | $1.90 | $0.1 | Imported from OpenRouter | openrouter.ai |
| 11 Jun 2026 | $0.2 | $0.88 | $0.11 | Imported from OpenRouter | openrouter.ai |