Skip to main content
The plugin may not be enabled. To fix this:
1

Open the Plugins window

Go to Edit → Plugins in the Unreal Editor menu bar.
2

Search for Node to Code

Type “Node to Code” in the search box and verify the checkbox is enabled.
3

Restart the editor

Click Restart Now when prompted. The toolbar button will appear after the restart.
Node to Code only supports Win64 and Mac platforms. It will not load on Linux or console targets.
Check the following in order:
  1. Verify your API key — Go to Edit → Project Settings → Node to Code → LLM Services → [Provider] → API Key and confirm the key is entered correctly with no extra spaces.
  2. Verify the correct provider is selected — The active provider in LLM Provider must match the API key you entered.
  3. Check your internet connection — Cloud providers (OpenAI, Anthropic, Gemini, DeepSeek) require an active internet connection.
  4. Check your API account — Ensure your account has credits and the selected model is available on your plan.
For more detail, set Logging → Min Severity to Info and check the Output Log in the Unreal Editor for the full error message.
This can happen when the Blueprint is very complex or the model has trouble with the schema. Try these steps:
  1. Reduce translation depth — Set Translation Depth to 0 (default). Higher depths send significantly more tokens and can exceed context windows.
  2. Test with a simpler Blueprint — Try a small function with just a few nodes to confirm the pipeline works.
  3. Switch to a more capable model — For complex Blueprints, try Claude 4 Sonnet or Gemini 2.5 Pro instead of a smaller/cheaper model.
  4. Check for unsupported node types — Uncommon node types may produce unexpected output. See Supported Node Types.
LLM responses can take time, especially for large Blueprints or reasoning models. The request timeout is set to 3600 seconds (1 hour) to accommodate complex translations.To speed things up:
  • Reduce translation depth — Each additional depth level can multiply token count.
  • Select fewer nodes — Translate a subset of your Blueprint at a time rather than the entire graph.
  • Use a faster model — Gemini 2.5 Flash and o4 Mini are optimized for speed and cost without sacrificing quality.
  • Remove large reference files — Large reference source files increase prompt size. Check Estimated Reference File Tokens in settings.
Several settings directly impact cost:
  • Translation Depth — Each level multiplies the amount of Blueprint data sent. Keep at 0 unless you specifically need nested translation.
  • Reference Source Files — Large .h/.cpp files add significant tokens. The Estimated Reference File Tokens field in settings shows the current total. Remove files you don’t need.
  • Model selection — Reasoning models (o1, o3, DeepSeek R1) use more output tokens. For most translations, o4 Mini, Gemini 2.5 Flash, or Claude 4 Sonnet offer the best cost-to-quality ratio.
For zero-cost translation, use Ollama or LM Studio with a local model.
1

Confirm Ollama is running

Open a terminal and run ollama serve. If it is already running, you will see a message indicating the server is active.
2

Pull the model

Run ollama pull qwen3:32b (or whichever model you have configured) to ensure the model is downloaded locally.
3

Verify the endpoint

In Project Settings → Node to Code → LLM Services → Ollama, confirm the host and port match the Ollama server (default: http://localhost:11434).
4

Check firewall rules

If Ollama is running on a remote machine, ensure the port is reachable from the machine running Unreal Editor.
1

Open LM Studio and load a model

Launch LM Studio, go to the My Models tab, and load the model you want to use.
2

Start the local server

In LM Studio, navigate to the Local Server tab and click Start Server.
3

Verify the endpoint

The default server endpoint is http://localhost:1234. Confirm this matches the Server Endpoint setting in Project Settings → Node to Code → LLM Services → LM Studio.
4

Confirm the model name

Set the Model Name in plugin settings to exactly match the model identifier shown in LM Studio.
Node to Code supports Win64 and Mac only. This is defined in NodeToCode.uplugin:
"PlatformAllowList": ["Win64", "Mac"]
It will not load on Linux, Android, iOS, or console platforms. If you need to ship to those platforms, the plugin is editor-only and will not be included in packaged builds regardless.

​
Enabling detailed logging

To see full debug output in the Unreal Output Log:
  1. Go to Edit → Project Settings → Node to Code → Logging
  2. Set Min Severity to Info
  3. Reproduce the issue and check the Output Log (Window → Output Log) for [NodeToCode] entries

​
Getting more help

Discord Community

Ask questions and get help from the community and developer

GitHub Issues

Report bugs or request features on GitHub

Build docs developers (and LLMs) love

Get started for freeTalk to us