Skip to main content
Convert a trained model into a deployable format: GGUF (quantized), MLX (Apple Silicon), MNN (Alibaba mobile), or QNN (Qualcomm). The export runs asynchronously; poll GET /v1/exports/:id to track completion.

Method + Path

Authentication

Include your PIN API token or JWT in the Authorization header as a Bearer token.

Path parameters

string
required
The model identifier (e.g., md_abc123def456).

Request body

string
required
Target export format. Must be one of: gguf, mlx, mnn, qnn.

Request body example

Request example

Response

Returns a standard PIN success envelope with the export task.
integer
required
0 on success, nonzero on error.
object
required
string
Human-readable message (empty on success).

Example response

Export process

  1. If the model has no merged artifact, the LoRA adapter is merged into the base weights first.
  2. The converter tool runs in a background goroutine:
    • GGUF: Converts to FP16 then quantizes to Q4_K_M using llama.cpp.
    • MLX: Runs mlx_lm.convert or the tool configured by ARIAPIN_CONVERT_MLX.
    • MNN: Runs mnnconvert or the tool configured by ARIAPIN_CONVERT_MNN.
    • QNN: Runs qairt-converter or the tool configured by ARIAPIN_CONVERT_QNN.
  3. On success, the artifact path is registered and the model status returns to ready.

Errors