LM-Kit OneDocs2026.8.10lm-kit.com
Infrastructure

Model Management

List running models.#

GET/lmkit/v1/models/running

Returns a list of all models currently loaded in memory, including their size, reference count, and last usage time.

Responses

StatusTypeDescription
200[]

OK

curl -X GET "$LMKIT_ONE_URL/lmkit/v1/models/running" \
  -H "Authorization: Bearer $LMKIT_API_KEY"

Load a model into memory.#

POST/lmkit/v1/models/load

Pre-loads a model into memory so that subsequent inference requests do not incur the initial loading latency. If the model is already loaded, the endpoint returns immediately with an "AlreadyLoaded" status. The model will be automatically downloaded if it is not available locally.

Request body

application/json ·

PropertyTypeDescription
modelrequiredstring

The model identifier to load (e.g. "gemma3:4b").

Responses

StatusTypeDescription
200

OK

400

Bad Request

404

Not Found

500application/json

Internal Server Error

curl -X POST "$LMKIT_ONE_URL/lmkit/v1/models/load" \
  -H "Authorization: Bearer $LMKIT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": ""
}'

Unload a model from memory.#

POST/lmkit/v1/models/unload

Removes a model from the in-memory cache and frees its resources. A model that is currently in use (has active inference requests) cannot be unloaded and will return a 409 Conflict status.

Request body

application/json ·

PropertyTypeDescription
modelrequiredstring

The model identifier to unload (e.g. "gemma3:4b").

Responses

StatusTypeDescription
200

OK

400

Bad Request

404

Not Found

409

Conflict

500application/json

Internal Server Error

curl -X POST "$LMKIT_ONE_URL/lmkit/v1/models/unload" \
  -H "Authorization: Bearer $LMKIT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": ""
}'

List all available models in the catalog.#

GET/lmkit/v1/models/catalog

Returns the complete catalog of predefined models with their metadata, including the functional type(s) of each model, download status, and file size. Use the optional 'type' query parameter to return only models that expose a given type.

Parameters

NameInTypeDescription
typequery

Optional filter by model type. If provided, only models that expose the specified type are returned.

Responses

StatusTypeDescription
200[]

OK

curl -X GET "$LMKIT_ONE_URL/lmkit/v1/models/catalog" \
  -H "Authorization: Bearer $LMKIT_API_KEY"

Download a model.#

POST/lmkit/v1/models/pull

Downloads a model from the remote repository to local storage. If the model is already downloaded, returns immediately with an "AlreadyAvailable" status. The download runs synchronously; the response is sent when the download completes.

Request body

application/json ·

PropertyTypeDescription
modelrequiredstring

The model identifier to load (e.g. "gemma3:4b").

Responses

StatusTypeDescription
200

OK

400

Bad Request

404

Not Found

500application/json

Internal Server Error

curl -X POST "$LMKIT_ONE_URL/lmkit/v1/models/pull" \
  -H "Authorization: Bearer $LMKIT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": ""
}'