Qwen: Qwen3.5-35B-A3B Qwen

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...

Architecture

  • Modality: text+image+video->text
  • InputModalities: text, image, video
  • OutputModalities: text
  • Tokenizer: Qwen3

ContextAndLimits

  • ContextLength: 262144 Tokens
  • MaxResponseTokens: 65536 Tokens
  • Moderation: Disabled

Pricing

  • Prompt1KTokens: 1.625e-07 ₽
  • Completion1KTokens: 1.3e-06 ₽
  • InternalReasoning: ₽
  • Request: ₽
  • Image: ₽
  • WebSearch: ₽