@@ -770,9 +770,9 @@ To add specific models to the configuration, configure as such:
770770
771771### Local Models
772772
773−Local models can also be configured via OpenAI-compatible API. Here are two common examples:
774−
775−#### Ollama
773+Crush can auto-discovers models from local providers. Add a custom provider
774+with `type` set to `omlx`, `lmstudio`, `litellm`, or `ollama` and leave out
775+the models list. Crush will populate the model list automatically.
776776
777777```json
778778{
@@ -780,37 +780,36 @@ Local models can also be configured via OpenAI-compatible API. Here are two comm
780780 "ollama": {
781781 "name": "Ollama",
782782 "base_url": "http://localhost:11434/v1/",
783− "type": "openai-compat",
784− "models": [
785− {
786− "name": "Qwen 3 30B",
787− "id": "qwen3:30b",
788− "context_window": 256000,
789− "default_max_tokens": 20000
790− }
791− ]
783+ "type": "ollama"
792784 }
793785 }
794786}
795787```
796788
797−#### LM Studio
789+#### Manual Model Configuration
790+
791+You can still list models explicitly. User-defined models always take
792+precedence over discovered ones, and any fields you set won't be overwritten
793+by auto-discovery. Auto discovery will run if the model list is empty for any
794+`openai-compat` provider or if you pass `"auto_discovery": true` it will merge
795+ the found models with your hand configured ones.
798796
799797```json
800798{
801799 "providers": {
802− "lmstudio": {
803− "name": "LM Studio",
804− "base_url": "http://localhost:1234/v1/",
805− "type": "openai-compat",
800+ "ollama": {
801+ "name": "Ollama",
802+ "base_url": "http://localhost:11434/v1/",
803+ "type": "ollama",
806804 "models": [
807805 {
808806 "name": "Qwen 3 30B",
809− "id": "qwen/qwen3-30b-a3b-2507",
807+ "id": "qwen3:30b",
810808 "context_window": 256000,
811809 "default_max_tokens": 20000
812810 }
813− ]
811+ ],
812+ "auto_discovery": true
814813 }
815814 }
816815}