ComfyUI Node
Qwen Thinking Prompt (Before KSampler)
A ComfyUI node in Qwen/Thinking with 8 inputs and 2 outputs.
Qwen Thinking Prompt (Before KSampler)
- model
- tokenizer
- refined_prompt
- thinking
◄user_prompt—►
◄prompt_body<think>
{instructions}
User prompt:
{user_prompt}
Final prompt:
</think>
Final prompt:►
◄instruction_bodyYou are an expert Stable Diffusion prompt engineer.
Rules:
- Preserve the exact original meaning and intent.
- Do NOT add any new concepts, styles, subjects, or details not present or implied.
- Remove redundancy and make the prompt concise but descriptive.
- Structure as a single, clean, comma-separated line.
- Output ONLY the final prompt. No reasoning, no extra text.►
◄max_new_tokens256►
◄temperature0.60►
◄top_p0.85►
CategoryQwen/Thinking
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| model | QWEN_THINKING_MODEL | — | |
| tokenizer | QWEN_TOKENIZER | — | |
| user_prompt | STRING | — | |
| prompt_body | STRING | <think> {instructions} User prompt: {user_prompt} Final prompt: </think> Final prompt: | — |
| instruction_body | STRING | You are an expert Stable Diffusion prompt engineer. Rules: - Preserve the exact original meaning and intent. - Do NOT add any new concepts, styles, subjects, or details not present or implied. - Remove redundancy and make the prompt concise but descriptive. - Structure as a single, clean, comma-separated line. - Output ONLY the final prompt. No reasoning, no extra text. | — |
| max_new_tokens | INT | 25632–512 | — |
| temperature | FLOAT | 0.600.1–1.2 | — |
| top_p | FLOAT | 0.850.1–1 | — |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| refined_prompt | STRING | — |
| thinking | STRING | — |