Nodes/ComfyUI-Qwen/Qwen Generation 🐼
ComfyUI Node

Qwen Generation 🐼

A ComfyUI node in 🐼Qwen with 6 inputs and 1 output.

By SXQBW·Created about a year ago·Updated about a year ago· 28
Qwen Generation 🐼
    • text
    model_nameQwen3-4B
    quantization👍 4-bit (VRAM-friendly)
    enable_thinkingfalse
    promptThis is the prompt text used for generating images with Fulx: "In the style of GHIBSKY, a cyberpunk panda holding a neon sign that reads: 'Designed by SXQBW'". Please optimize, supplement and improve the prompt text according to its content, and make the generated image effect the best.Just provide the best answer content.
    max_new_tokens2048
    unload_after_generationtrue
    Category🐼Qwen

    Inputs (6)

    NameTypeDefaultDescription
    model_nameCOMBOQwen3-4B选择可用的模型版本。 | Select the available model version.
    quantizationCOMBO👍 4-bit (VRAM-friendly)选择量化级别: ✅ 4-bit: 显著减少显存使用。 ⚖️ 8-bit: 平衡精度和性能。 🚫 None: 使用原始精度(需要高端GPU)。 Select the quantization level: ✅ 4-bit: Significantly reduces VRAM usage. ⚖️ 8-bit: Balances precision and performance. 🚫 None: Uses original precision (requires high-end GPU).
    enable_thinkingBOOLEANfalse启用或禁用思考模式。思考模式适用于复杂推理任务,非思考模式适用于高效对话。 | Enable or disable thinking mode. Thinking mode is suitable for complex reasoning tasks, while non-thinking mode is optimized for efficient conversations.
    promptSTRINGThis is the prompt text used for generating images with Fulx: "In the style of GHIBSKY, a cyberpunk panda holding a neon sign that reads: 'Designed by SXQBW'". Please optimize, supplement and improve the prompt text according to its content, and make the generated image effect the best.Just provide the best answer content.输入提示文本 | Enter the prompt text
    max_new_tokensINT204864–6144控制生成的最大token数 | Control the maximum number of tokens to generate
    unload_after_generationBOOLEANtrue生成后是否卸载模型以释放资源。启用此选项可减少内存占用,但会增加下次使用时的加载时间。 | Whether to unload the model after generation to free up resources. Enabling this option reduces memory usage but increases load time for subsequent uses.

    Outputs (1)

    NameTypeDescription
    textSTRING