Skip to main content
Qwen (通义千问) is Alibaba Group’s family of large language models. With a wide range of models from ultra-fast flash variants to the flagship Qwen Max, Qwen covers a broad spectrum of capability and cost efficiency. Qwen models are particularly strong in Chinese, Japanese, Korean, and other Asian languages, as well as mathematics and coding.

Available Models

ModelTypeStrengthsPlan Required
Qwen MaxTextFlagship — most capable Qwen model, strong reasoningPro+ (Premium)
Qwen PlusTextBalanced quality and speedPro+
Qwen TurboTextFast, efficient standard modelStandard (all plans)
Qwen FlashTextUltra-fast responses, lightweight tasksStandard (all plans)
Qwen3 variantsTextLatest generation models (various sizes)Premium or Standard
Qwen ImageImageText-to-image generationPro+
Qwen Image EditImageImage editing and modificationPro+

Strengths

Multilingual (especially Asian languages): Qwen models are trained with extensive multilingual data, with particularly strong performance in Chinese (Simplified and Traditional), Japanese, and Korean. For tasks in these languages, Qwen often significantly outperforms Western-focused models. Mathematics: Qwen Max and Qwen3 variants perform strongly on mathematical reasoning, including algebra, calculus, statistics, and quantitative analysis. Coding: Qwen models have strong coding capabilities across many programming languages, with particular strength in tasks requiring code + Chinese-language documentation or explanation. Cost-effective high-volume use: Qwen Turbo and Qwen Flash offer strong capability at standard model efficiency — ideal for high-volume use cases where premium quota conservation matters. Image generation and editing: Qwen provides image generation (Qwen Image) and image editing (Qwen Image Edit) models in addition to text models.

Best Use Cases

Chinese language tasks

Writing, translation, summarization, and analysis in Chinese. Qwen Max produces natural, nuanced Chinese text.

Multilingual applications

Japanese, Korean, and other Asian language tasks where most Western models struggle.

Mathematics

Quantitative problem-solving, statistics, algebra, and mathematical reasoning.

Cost-efficient high volume

Qwen Turbo and Flash for high-volume standard model tasks without using premium quota.

Qwen Image Models

In addition to text generation, ZeroTwo includes Qwen’s image models: Qwen Image: Text-to-image generation — describe an image and Qwen generates it. Available in Studio at /studio/images. Qwen Image Edit: Image modification — upload an existing image and describe the changes you want. Useful for iterative image refinement. Both image models require Pro+ plan access.

Choosing a Qwen Model

If you need…Use…
Best overall Qwen capabilityQwen Max
Balanced quality / speedQwen Plus
Fast standard model, everyday tasksQwen Turbo
Ultra-fast lightweight responsesQwen Flash
Latest generation capabilityQwen3 variants
Text-to-image generationQwen Image
Image editingQwen Image Edit

For Chinese-language work, try Qwen Max first — it consistently produces high-quality, natural Chinese text that closely matches how a native speaker would phrase things.