ComfyUI Node: ThinkingLLM Prompt Enhancer
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.
Category
ThinkingLLM
Inputs
model_name
- Qwen3.5-4B-heretic-v2 [DL: 7.5GB, VRAM: 9.0GB]
- Qwen3.5-9B-ultra-uncensored-heretic-v2 [DL: 17GB, VRAM: 19.0GB]
- Qwen3.5-4B [DL: 7.5GB, VRAM: 9.0GB]
- Qwen3.5-9B [DL: 17GB, VRAM: 19.0GB]
- Qwen3-VL-4B-Instruct-Abliterated [DL: 7.5GB, VRAM: 6.0GB]
- Qwen3-VL-8B-Instruct-Abliterated [DL: 15GB, VRAM: 12.0GB]
- Qwen3-VL-4B-Instruct-Unredacted-MAX [DL: 7.5GB, VRAM: 6.0GB]
- Qwen3-VL-8B-Instruct-Unredacted-MAX [DL: 15GB, VRAM: 12.0GB]
- Qwen3-VL-8B-Instruct-abliterated-v1 [DL: 15GB, VRAM: 12.0GB]
- Qwen3-VL-4B-Instruct-abliterated-v1 [DL: 7.5GB, VRAM: 6.0GB]
- Qwen3-VL-32B-Heretic-v2 [DL: 60GB, VRAM: 48.0GB]
- Gemma-4-26B-A4B-it [DL: 51.6GB, VRAM: 42.0GB]
- Gemma-4-E4B-it [DL: 16GB, VRAM: 7.0GB]
- Gemma-4-E2B-it [DL: 10.3GB, VRAM: 4.0GB]
- Gemma-4-31B-it [DL: 62.6GB, VRAM: 48.0GB]
- Gemma-4-12B [DL: 24GB, VRAM: 24.0GB]
- Gemma-4-E4B-Uncensored [DL: 16GB, VRAM: 7.0GB]
- Gemma-4-E2B-Uncensored [DL: 10.3GB, VRAM: 4.0GB]
- Gemma-4-26B-A4B-Heretic [DL: 51.6GB, VRAM: 42.0GB]
- Gemma-4-E4B-Uncensored-Aggressive [DL: 16GB, VRAM: 7.0GB]
- Gemma-4-E2B-Uncensored-Aggressive [DL: 10.3GB, VRAM: 4.0GB]
- qwen3-4b-Z-Image-Engineer [DL: 7.5GB, VRAM: 9.0GB]
- Qwen3.5-4B-heretic-v2 [DL: 7.5GB, VRAM: 8.0GB]
- Qwen3.5-9B-ultra-uncensored-heretic-v2 [DL: 17GB, VRAM: 18.0GB]
- Qwen3.5-4B [DL: 7.5GB, VRAM: 8.0GB]
- Qwen3.5-9B [DL: 17GB, VRAM: 18.0GB]
quantization
- 4-bit (VRAM-friendly)
- 8-bit (Balanced)
- None (FP16)
attention_mode
- auto
- sage
- flash_attention_2
- sdpa
use_torch_compile BOOLEAN
device
- auto
- cuda
- cpu
- mps
prompt_text STRING
enhancement_style
- ✍️ Custom Only (no preset)
- 📖 LTX 2.3 NSFW T2V Scene
- 📖 Wan 2.2 NSFW T2V Scene (20s)
- 🍿 Wan 2.2 NSFW T2V Timeline (3)
- 🍿 Wan 2.2 NSFW T2V Timeline (5s)
- 🎬 Wan 2.2 NSFW T2V Timeline (20s)
- 🎥 Wan 2.2 NSFW T2V Scene (3s)
- 🎥 Wan 2.2 NSFW T2V Scene (5s)
- 📝 Enhance
- 📝 Refine
- 📝 Creative Rewrite
- 📝 Detailed Visual
- 📝 Artistic Style
- 📝 Technical Specs
custom_system_prompt STRING
max_tokens INT
temperature FLOAT
top_p FLOAT
repetition_penalty FLOAT
keep_model_loaded BOOLEAN
seed INT
keep_last_prompt BOOLEAN
stream_tokens_to_terminal BOOLEAN
enable_thinking BOOLEAN
hf_token STRING
Outputs
STRING
STRING
Extension: ComfyUI-ThinkingLLM
A multimodal ComfyUI AI node with Qwen3.5, Qwen3-VL, Qwen2.5-VL, Qwen3, and Gemma 4 integrations. Features live thinking in the terminal to see what the LLM is doing in real time.
Authored by goodguy1963
Looking for a different node?
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.