ComfyUI Node
🔌 Provider Loader
🔌 FiL Provider Loader — selects an LLM provider, model, and generation parameters. Outputs a `config` dict for downstream FiL nodes.
🔌 Provider Loader
- config
- model
◄providerollama►
◄model(loading...)►
◄refresh_modelsfalse►
◄temperature0.70►
◄max_tokens0►
◄rate_limit_ms100►
◄max_image_side1024►
Category🎨 FiL Design/LLM
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| provider | COMBO | ollama | Choose where the model runs. Local for offline work, cloud for hosted models. |
| model | COMBO | (loading...) | Choose which model to use. If the list is empty, refresh it or check the provider account. |
| refresh_models | BOOLEAN | false | Reload the model list. Use after adding a new model or API key. |
| temperature | FLOAT | 0.700–2 | Sampling temperature — higher is more creative, lower is more deterministic. |
| max_tokens | INT | 00–65536 | Maximum tokens in the response. 0 = provider default (no explicit limit). |
| rate_limit_ms | INT | 1000–5000 | Minimum pause between requests, in milliseconds. On a 429 the pack widens this by itself for a couple of minutes, honouring the provider's own Retry-After when it sends one. |
| max_image_side | INT | 1024128–4096 | Images are downscaled so their longest side does not exceed this value. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| config | FIL_PROVIDER_CONFIG | Provider configuration dict for FiL nodes. |
| model | STRING | Normalised model name string. |