SubtitleEdit/subtitleedit · error · FileNotFoundException

llama.cpp model not found - please download a model first.

Error message

llama.cpp model not found - please download a model first.

What it means

Thrown by LlamaCppServerManager when starting llama-server: the resolved modelPath (a .gguf file) does not exist on disk. It is a FileNotFoundException whose FileName is the missing path. The check happens AFTER the executable presence check, so reaching it means llama-server itself is present but the model weight file is not.

Source

Thrown at src/libuilogic/LlamaCpp/LlamaCppServerManager.cs:506

                Configuration.Settings.Tools.LlamaCppApiUrl = ApiUrl;
                return;
            }

            // Server not running, or running with a different model - (re)start.
            if (_serverProcess != null)
            {
                StopServerInternal();
            }

            var exe = GetExecutable();
            if (!File.Exists(exe))
            {
                throw new FileNotFoundException("llama-server executable not found - please download llama.cpp first.", exe);
            }

            if (!File.Exists(modelPath))
            {
                throw new FileNotFoundException("llama.cpp model not found - please download a model first.", modelPath);
            }

            string? mmprojPath = null;
            if (model.MmprojFileName != null)
            {
                mmprojPath = GetModelPath(model.MmprojFileName);
                if (!File.Exists(mmprojPath))
                {
                    throw new FileNotFoundException("llama.cpp vision projector not found - please download the model first.", mmprojPath);
                }
            }

            var port = FindFreeLoopbackPort();
            var psi = new ProcessStartInfo
            {
                WorkingDirectory = Path.GetDirectoryName(exe) ?? GetAndCreateFolder(),
                FileName = exe,
                UseShellExecute = false,

View on GitHub (pinned to 17a9f07487)

Solutions

1. Re-download the model from Subtitle Edit (Auto-translate > llama.cpp > Download) so the .gguf lands in GetAndCreateModelsFolder().
2. Verify the resolved modelPath with File.Exists before calling StartServer and surface a user-facing message if it is missing.
3. If you keep models elsewhere, pass an absolute path via --translate-model:<path.gguf> so resolution skips the models folder.
4. Check for a half-written download (size far smaller than expected) and delete it so SE re-downloads.

Example fix

// before
if (!File.Exists(modelPath))
    throw new FileNotFoundException("llama.cpp model not found - please download a model first.", modelPath);

// after (caller-side guard)
if (!File.Exists(modelPath))
    return Result.Fail($"Model not found at '{modelPath}'. Download it via SE Auto-translate > llama.cpp, or pass --translate-model.");

public static void EnsureModelPresent(string modelPath)
{
    if (!File.Exists(modelPath))
        throw new InvalidOperationException(
            $"Model '{modelPath}' is missing. Download via SE Auto-translate > llama.cpp.");
    var info = new FileInfo(modelPath);
    if (info.Length < 1024 * 1024)
        throw new InvalidOperationException($"Model '{modelPath}' looks truncated ({info.Length} bytes).");
}

null

null

When it happens

Trigger: Calling the server start path with a LlamaCppModel whose local file was deleted, moved, or never finished downloading; GetModelPath() resolving to a path under the SE models folder that no longer contains the .gguf.

Common situations: User picked a model in the SE GUI, then cleaned the Models folder; an interrupted download left a partial file that File.Exists rejects; antivirus quarantined the .gguf; the model file lives under %AppData% on a different user account.

Related errors

• Translate model '{name}' not found in {LlamaCppServerManager
• llama-server executable not found - please download llama.cp
• llama.cpp vision projector not found - please download the m
• Translate model file not found: {name}
• No llama.cpp translate model found in {LlamaCppServerManager

AI-assisted analysis of SubtitleEdit/subtitleedit@17a9f07487 (2026-08-13). Data as JSON: /api/errors/839b0a584cee8081. Report an issue: GitHub.