Qwen3.5-Flash
by AlibabaRank #175Score 68.6
The Qwen3.5 native vision-language Flash models are built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. Compared to the...
性能概览
评分
68.6
排名
#175
24小时变化
+1
7天变化
-8
状态
stable置信度
high信号评分
| 信号 | 标准化 | 权重 | 贡献 | 新鲜度 |
|---|---|---|---|---|
Capabilities capability | 83.3 | 20% | 16.7 | 2026-09-22T15:16:39.103Z |
Benchmarks benchmark | 66.3 | 30% | 19.9 | 2026-09-22T15:16:39.103Z |
Pricing pricing_tier | 99.7 | 15% | 15.0 | 2026-09-22T15:16:39.103Z |
Context Window context_window | 95.2 | 10% | 9.5 | 2026-09-22T15:16:39.103Z |
Recency recency | 94.8 | 15% | 14.2 | 2026-09-22T15:16:39.103Z |
Output Capacity output_capacity | 77.0 | 10% | 7.7 | 2026-09-22T15:16:39.103Z |
主要驱动因素
positive
Benchmarks
positive
Capabilities
Supports reasoning, vision, tools, JSON mode, streaming
5/7
positive
Pricing
$0.26/M output tokens
$0.26
positive
Recency
Released over 1 year ago
7mo ago
功能
| 功能 | 支持 |
|---|---|
| 视觉 | 是 |
| 推理 | 是 |
| JSON模式 | 是 |
| 流式输出 | 是 |
| 函数调用 | 是 |
| 网页搜索 | 否 |
价格
输入/百万令牌
$0.07
输出/百万令牌
$0.26
上下文窗口
1000K
最大输出
66K