qwentext + image + video → text
Qwen: Qwen3.5-Flash
Canonical ID: qwen/qwen3.5-flash-02-23
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...
Artificial Analysis Benchmarks
VerifiedNo standardized Artificial Analysis benchmark scores recorded for this variant yet.
Token Pricing & Economics
Prompt Input (1M Tokens)
Input$0.07
Completion Output (1M Tokens)
Output$0.26
Prompt KV-Cache Read (1M Tokens)
Cached$0.00
Context Window Capacity
Context976.563k tokens
Supported Parameters & Features
frequency_penaltyinclude_reasoningmax_tokenspresence_penaltyreasoningresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_p