← All models View on HuggingFace →
Qwen: Qwen3.5-122B-A10B
qwen/qwen3.5-122b-a10b
VisionTool useJSONReasoningStreaming
The Qwen3.5 122B-A10B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. In terms of...
Pricing
Input
$0.26 / 1M
Output
$2.08 / 1M
Specs
Context
262,144 tokens
Input
text, image, video
Output
text
Released: 2026-02
Supported parameters
frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Open weights · HuggingFace
868,631 downloads/mo
558 likes
apache-2.0 image-text-to-text