Zackriya-Solutions/meetily · error

Failed to load model '{}': {}

Error message

Failed to load model '{}': {}

What it means

`WhisperEngine::load_model(target_model)` failed while preparing the engine for import. target_model comes from the import request's model argument or the transcript_settings row. Typical inner causes: the named ggml model file is absent from the models directory (the discover_models warning just above is non-fatal, so discovery problems flow into this error), the file is corrupt/truncated, or the GPU backend (Metal/CUDA/Vulkan per build features) fails to initialize.

Source

Thrown at frontend/src-tauri/src/audio/import.rs:787

            let current_model = e.get_current_model().await;
            let needs_load = match &current_model {
                Some(loaded) => loaded != &target_model,
                None => true,
            };

            if needs_load {
                info!(
                    "Loading Whisper model '{}' (current: {:?})",
                    target_model, current_model
                );

                if let Err(e) = e.discover_models().await {
                    warn!("Model discovery error (continuing): {}", e);
                }

                e.load_model(&target_model)
                    .await
                    .map_err(|e| anyhow!("Failed to load model '{}': {}", target_model, e))?;
            }

            Ok(e)
        }
        None => Err(anyhow!("Whisper engine not initialized")),
    }
}

/// Get or initialize the Parakeet engine
async fn get_or_init_parakeet<R: Runtime>(
    app: &AppHandle<R>,
    requested_model: Option<&str>,
) -> Result<Arc<ParakeetEngine>> {
    use crate::parakeet_engine::commands::PARAKEET_ENGINE;

    let engine = {
        let guard = PARAKEET_ENGINE.lock().unwrap_or_else(|e| e.into_inner());
        guard.as_ref().cloned()

View on GitHub (pinned to 0281737d87)

Solutions

1. List the models the app can actually see (model-list command or discover_models result) and confirm target_model is present before reimporting
2. Download the correct ggml file into the models dir: frontend/models in dev; Application Support/Meetily/models on macOS, %APPDATA%\Meetily\models on Windows
3. Omit the model argument so get_configured_model falls back to the configured/default model, or pick one known to exist
4. If the inner error mentions Metal/CUDA/Vulkan, retry on a CPU-feature build or fix the GPU runtime

Example fix

// before — load and hope
e.load_model(&target_model).await
    .map_err(|e| anyhow!("Failed to load model '{}': {}", target_model, e))?;

// after — verify availability first (adapt to discover_models' return type)
if let Ok(models) = e.discover_models().await {
    if !models.iter().any(|m| m == &target_model) {
        return Err(anyhow!("Model '{}' not found. Available: {:?}", target_model, models));
    }
}
e.load_model(&target_model).await
    .map_err(|e| anyhow!("Failed to load model '{}': {}", target_model, e))?;

// verify the ggml file exists before import
let models_dir = app_models_dir(); // dev: frontend/models, prod: app-support/Meetily/models
let expected = models_dir.join(format!("ggml-{}.bin", target_model));
if !expected.exists() {
    return Err(anyhow!("Model '{}' not downloaded: {}", target_model, expected.display()));
}

Catch the load error and fall back to a known-present model (or omit the model argument to use the configured default) before giving up — the inner error text distinguishes 'not found' from GPU-init failures, which determines whether a CPU retry helps.

When it happens

Trigger: Importing with a model name (request arg or settings row) that was never downloaded; the models directory relocated or cleaned; a partially downloaded ggml file; a GPU-feature build on a machine whose GPU runtime is missing.

Common situations: User picked 'large-v3' without ever downloading it; app support dir cleaned between sessions; settings restored from another machine without the model files; GPU driver update broke Metal/CUDA init.

Related errors

• Failed to load model {}: {}
• Whisper transcription failed on segment {}: {}
• Whisper engine not initialized
• Failed to load Whisper model '{}': {}
• Failed to load Parakeet model {}: {}

AI-assisted analysis of Zackriya-Solutions/meetily@0281737d87 (2026-08-16). Data as JSON: /api/errors/46cd8ae7b560f877. Report an issue: GitHub.