mudler/LocalAI · error

error: model not loaded

Error message

error: model not loaded

What it means

A transcription request arrived before the voxtral model was loaded. Guard: track loaded state and reject inference requests until LoadModel succeeds.

Source

Thrown at backend/go/voxtral/csrc/govoxtral.c:39

#ifdef USE_METAL
    if (!metal_initialized) {
        vox_metal_init();
        metal_initialized = 1;
    }
#endif

    ctx = vox_load(model_dir);
    if (ctx == NULL) {
        fprintf(stderr, "error: failed to load voxtral model from %s\n", model_dir);
        return 1;
    }

    return 0;
}

const char *transcribe(const char *wav_path) {
    if (ctx == NULL) {
        fprintf(stderr, "error: model not loaded\n");
        return "";
    }

    if (last_result != NULL) {
        free(last_result);
        last_result = NULL;
    }

    last_result = vox_transcribe(ctx, wav_path);
    if (last_result == NULL) {
        fprintf(stderr, "error: transcription failed for %s\n", wav_path);
        return "";
    }

    return last_result;
}

void free_result(void) {

View on GitHub (pinned to 44413a9d06)

Solutions

1. Load the model first (LoadModel) before requesting transcription.
2. Check earlier logs for a failed model load.

When it happens

Trigger: Thrown at backend/go/voxtral/csrc/govoxtral.c:39 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of mudler/LocalAI@44413a9d06 (2026-08-15). Data as JSON: /api/errors/49a594abadd09b81. Report an issue: GitHub.