This workflow requires my Comfyui-lmstudio-prompt-enhancer custom node which is installable via Comfy Manager.

Learn more about this custom node

🛈 IMPORTANT!!!

You will need to search for, download, and load the model google/gemma-4-12b-qat in LM studio. You will also need to start the API server and generate an API Key. Learn how to configure LM Studio here.

Notes for using google/gemma-4-12b-qat

• I chose the QAT variant over the non-QAT variant as this workflow can benefit from optimized memory utilization.

• This is a memory efficient model which is fast and does a decent job at generating aesthetically pleasing prompts.

• I set the max tokens to 1500, this higher value is to accommodate the reasoning tokens.

• If the image looks flat, check the logs in LM Studio, the last log entry will contain a JSON path x.choices[0].message.content if the content field is empty that means the LLM ran out of reasoning tokens, this can be resolved by increasing max_tokens.