Zackriya-Solutions/meetily · error

Unknown model: {}

Error message

Unknown model: {}

What it means

models::get_model_by_name performs an exact string match against the hardcoded registry; None produces this error. Valid names are 'qwen3.5:2b', 'qwen3.5:4b', 'gemma3:4b', 'gemma3:1b' — the ':size' suffix is part of the id. Any casing difference, whitespace, or a name from another engine's list fails.

Source

Thrown at frontend/src-tauri/src/summary/summary_engine/client.rs:152

    app_data_dir: &PathBuf,
    model_name: &str,
    system_prompt: &str,
    user_prompt: &str,
    cancellation_token: Option<&CancellationToken>,
) -> Result<String> {
    // Check cancellation at start
    if let Some(token) = cancellation_token {
        if token.is_cancelled() {
            return Err(anyhow!("Generation cancelled before starting"));
        }
    }

    log::info!("Built-in AI generation request");
    log::info!("Model: {}", model_name);

    // Get model definition
    let model_def = models::get_model_by_name(model_name)
        .ok_or_else(|| anyhow!("Unknown model: {}", model_name))?;

    // Resolve model path with caching (avoids repeated filesystem I/O)
    let model_path = get_cached_model_path(app_data_dir, model_name)?;

    // Apply model-specific chat template
    let formatted_prompt =
        models::format_prompt(&model_def.template, system_prompt, user_prompt)?;
    // Get or initialize sidecar manager
    let manager = {
        let mut global_manager = SIDECAR_MANAGER.lock().await;
        if global_manager.is_none() {
            log::info!("Initializing sidecar manager");
            let new_manager = SidecarManager::new(app_data_dir.clone())?;
            *global_manager = Some(Arc::new(new_manager));
        }
        global_manager.clone().unwrap()
    };

View on GitHub (pinned to 0281737d87)

Solutions

1. Use an exact registry name including the size suffix, e.g. 'gemma3:1b'
2. Populate the model picker from get_available_models()/the list-models command instead of hardcoding names in the frontend
3. Validate the stored preference at startup and fall back to get_default_model() when it is unknown
4. Trim input before passing model names across the Tauri boundary

Example fix

// caller: normalize + validate before generating
let model_name = model_name.trim();
let model_def = models::get_model_by_name(model_name)
    .unwrap_or_else(models::get_default_model);
let text = client::generate_builtin(&app_data_dir, &model_def.name, sys, user, None).await?;

let model_name = input.trim();
if models::get_model_by_name(model_name).is_none() {
    return Err(anyhow!("Unknown model '{}' — valid: {:?}",
        model_name,
        models::get_available_models().iter().map(|m| m.name.clone()).collect::<Vec<_>>()));
}

// Narrow to a known model id before calling into the engine
fn known_model(name: &str) -> Option<&'static str> {
    const KNOWN: [&str; 4] = ["qwen3.5:2b", "qwen3.5:4b", "gemma3:4b", "gemma3:1b"];
    KNOWN.iter().find(|m| **m == name.trim()).copied()
}

When it happens

Trigger: Passing 'gemma3' instead of 'gemma3:4b'; passing a whisper/parakeet model id to the summary engine; a persisted settings value referencing a model removed from the registry in a newer app version.

Common situations: Stored user preference survives an app update that renamed models; frontend dropdown bound to the wrong field; copy-pasted ids with stray whitespace.

Related errors

• Model file not found: {}. Please download the model '{}' fir
• Unknown model: {}
• Generation failed: {}
• Sidecar error: {}
• Model '{}' not found

AI-assisted analysis of Zackriya-Solutions/meetily@0281737d87 (2026-08-16). Data as JSON: /api/errors/0c96cc0b9710a35c. Report an issue: GitHub.