The Qwen3.5 Series 35B-A3B is a native vision-language model designed with a hybrid architecture that integrates linear attention mechanisms and a sparse mixture-of-experts model, achieving higher inference efficiency. Its overall...
Rank
#117
Down 12% week over week
Tokens
18.9B
2026-08-14
Requests
2.4M
2,395,601
Tool calls
-
Input / M
$0.22
Output / M
$1.80
Cache / M
$0.22
Context
262K
262,144 tokens