Set MiMo LLM Service Connector 🐑

This node hooks Xiaomi's MiMo models into the MieNodes pack. It's the standard-tier MiMo connector - the one for a normal sk-... Open Platform key, not the Token Plan subscription (that's a separate node). Drop it, add your key, pick a MiMo model, and the rest of the pack's prompt tools can call it.

It ships in ComfyUI-MieNodes (ComfyUI_MieNodes), MieMieeeee's utility pack, under LLM Service Config.

Why you'd use it

Every prompt-generator node in this pack needs an LLM connection to do its work. This one produces a connection pointed at MiMo, Xiaomi's LLM line. If you already have MiMo credits - the standard Open Platform tier - this is how you spend them on prompt enhancement, translation, and the various generator nodes instead of piping everything through DeepSeek or SiliconFlow.

The MiMo lineup leans toward long-context and multimodal ("Omni") variants, which matters if you want the model to actually see a reference image while it rewrites a prompt. That's the workflow the KB's prompt-engineering doc frames as the modern default: an LLM reading your intent (and now your image) and emitting a clean instruction for an LLM-encoded diffusion model.

One thing worth knowing, from the pack's README: the MiMo connectors post to the OpenAI-compatible /v1/chat/completions endpoint but use max_completion_tokens rather than the legacy max_tokens, and they drop a few params (top_k, n, response_format) that the MiMo docs never show and would likely 400 on. That's handled for you inside the node - you don't have to think about it, but it's why this dedicated connector exists instead of just using the general one.

The inputs that matter

Required:

• api_token - your MiMo API key (sk-...). Optional if you keep it in the config file.
• model_select - dropdown of mimo-v2.5-pro (the default, deep-thinking, big context), mimo-v2.5 (full-modality Omni), mimo-v2-omni, mimo-v2-flash (low cost), mimo-v2-pro, or Custom. Reach for a -flash variant for cheap bulk prompt rewriting, -pro when quality matters, an Omni one if you're feeding it images.

Optional, shared with the pack's other connectors:

• custom_model - used only when model_select is Custom.
• config_file (default mie_llm_keys.json), config_key (default mimo), prefer_local_config (default true - the JSON file wins over the api_token field).

Output: llm_service_connector, which you wire into CallLLMService or any generator node.

Installing it

ComfyUI Manager → search ComfyUI-MieNodes → install → restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/MieMieeeee/ComfyUI-MieNodes

then restart. No model files - MiMo runs on Xiaomi's side. Nodes land under the 🐑 MieNodes menu.

Common issues

Standard vs Token Plan mixup. This node is for a standard sk-... key. If you're on Xiaomi's Token Plan / Coding Plan (keys look like tp-..., different base URL), you want SetMiMoTokenPlanLLMServiceConnector instead - the endpoints and billing differ.

Key precedence surprises. By default mie_llm_keys.json (the mimo entry) beats whatever you type in api_token. Keep the key in one place, or set prefer_local_config to false to force the node field.

Model errors after a Xiaomi rename. Switch model_select to Custom and paste the current id into custom_model.