Qwen3 VL 8B Thinking
Alibaba · Multimodal · Released Oct 2025
An 8-billion-parameter multimodal vision-language model from Alibaba that processes text, images, and video with extended reasoning capabilities and a 131k-token context window.
- Strengths
- Combines vision-language understanding with chain-of-thought reasoning to handle complex multimodal analysis and interpretation tasks.
- Best for
- Visual reasoning tasks requiring explanation of image or video content, where a smaller parameter count and faster inference are priorities.
- Limitations
- The 8B parameter count limits reasoning depth and multimodal understanding compared to larger Thinking variants like the 30B A3B Thinking or 235B A22B Thinking in the same lineup.
Input / 1M
$0.18
Output / 1M
$2.10
Cached input / 1M
—
Context window
131K
Price history
Input (solid)Output (dashed)
Price change
- 30d
- in — out —
- 90d
- in — out —
- 1y
- in increased 53.8% out increased 53.8%
- Since launch
- in increased 53.8% out increased 53.8%
Snapshots
| Effective | Input | Output | Cached in | Note | Source |
|---|---|---|---|---|---|
| 29 Jul 2026 | $0.18 | $2.10 | — | Imported from OpenRouter | openrouter.ai |
| 11 Jun 2026 | $0.117 | $1.36 | — | Imported from OpenRouter | openrouter.ai |