ComfyUI Node
Qwen 3.5 (GGUF)
A ComfyUI node in Qwen3.5 with 15 inputs and 2 outputs.
Qwen 3.5 (GGUF)
- image
- RESPONSE
- THINKING
◄modelQwen3.5-9B►
◄quantizationQ4_K_XL►
◄promptDescribe this image in detail.►
◄system_prompt►
◄max_tokens4096►
◄temperature0.70►
◄top_p0.80►
◄top_k20►
◄repeat_penalty1.00►
◄n_gpu_layers99►
◄ctx_size8192►
◄enable_thinkingfalse►
◄seed1►
◄cli_path►
CategoryQwen3.5
Inputs (15)
| Name | Type | Default | Description |
|---|---|---|---|
| model | COMBO | Qwen3.5-9B | Model size. 0.8B ~1GB, 2B ~2GB, 4B ~3GB, 9B ~6GB, 27B ~17GB (Q4) |
| quantization | COMBO | Q4_K_XL | GGUF quantization. XL = Unsloth Dynamic (smart mixed precision) |
| prompt | STRING | Describe this image in detail. | Text prompt for the model |
| system_prompt | STRING | Optional system prompt to set model behavior | |
| max_tokens | INT | 409664–32768 | Maximum tokens to generate |
| temperature | FLOAT | 0.700–2 | Sampling temperature (0.6-0.7 recommended for captioning) |
| top_p | FLOAT | 0.800–1 | Nucleus sampling threshold |
| top_k | INT | 201–100 | Top-K sampling |
| repeat_penalty | FLOAT | 1.000.5–2 | Penalty for repeated tokens |
| n_gpu_layers | INT | 99-1–200 | -1 or 99 offloads all layers to GPU |
| ctx_size | INT | 81921024–131072 | Context window size in tokens |
| enable_thinking | BOOLEAN | false | Enable thinking mode. Outputs reasoning in THINKING output. |
| seed | INT | 11–4294967295 | Random seed for reproducibility |
| imageopt | IMAGE | Image for vision tasks | |
| cli_pathopt | STRING | Path to llama-mtmd-cli binary. Auto-detected if empty. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| RESPONSE | STRING | — |
| THINKING | STRING | — |