Skip to main content

FAQ

Q: Which AI provider should I use?​

A: Start with xAI (Grok)—it's fast, cost-effective, and has strong Unreal Engine knowledge. Models like grok-code-fast-1 provide excellent value.

If you need vision support, try GPT-4o (OpenAI)—it's the industry standard for vision and has high-quality responses.

If you're analyzing huge Blueprints or need massive context, try Claude Opus 4 (Anthropic)—it has a 200k token context window, or Gemini Pro 1.5 (Google) with 1M+ tokens.

You can enable multiple providers and switch between them based on the task.

Q: How much do API calls cost?​

A: Costs vary by provider and model. Rough estimates (as of February 2026):

  • grok-code-fast-1: ~$0.50 per 1M tokens (very affordable)
  • GPT-4o: ~$5.00 per 1M tokens (moderate)
  • Claude Opus 4: ~$15.00 per 1M tokens (premium)
  • GPT-3.5 Turbo: ~$0.50 per 1M tokens (budget option)

Token usage examples:

  • Simple question: 200-500 tokens (fractions of a cent)
  • Blueprint analysis: 5,000-50,000 tokens ($0.02-$0.75 depending on model)
  • Large Blueprint with conversation context: 100,000+ tokens ($0.50-$1.50)

Tip: Use fast/cheap models for simple questions, save expensive models for complex analysis.

Check each provider's pricing page for current rates:

Q: Can I use Orion offline?​

A: In short, you can use Orion offline in limited capacity but not completely.

Anthropic, xAI, Google, and OpenAI require an internet connection to reach the provider APIs. However, Orion can still capture context while offline, which in turn can be used with the Epic Developer Assistant. As of Unreal Engine 5.7, the Epic Developer Assistant although limited at times, can work fully offline.

Future: The goal is to provide better integration with the Epic Developer Assistant to support more robust offline workflows, however for the moment the EDA does not support API or vision capabilities. This means Orion can't directly send screenshots or context. You can enable early EDA Integration and export Blueprint context to EDA (Settings > Enable EDA Integration), which will copy context to your clipboard, then paste into the EDA along with your query.

Q: Does Orion send my entire project to the AI?​

A: No. Orion only sends what you explicitly choose:

  • Blueprint analysis: Only the specific Blueprint you selected (via toolbar or right-click)
  • Describe to Orion: Only the asset you right-clicked
  • Screenshots: Only when you click the Screenshot button
  • Log lines: Only when "Include Last Log Lines" is enabled in settings

Nothing is sent without your action (clicking Send, choosing a right-click option, etc.).

Q: Where are my API keys stored?​

A: Encrypted in your user profile directory, never in project files or version control - Unless you choose to save there via custom path*.

*Available for custom needs, but changing the default save location is generally not recommended

Default Storage location

  • Windows: %AppData%\AzimuthOrion\

Keys are encrypted using platform-specific secure storage APIs. Even if someone accesses these files, they cannot read the keys without your user account access.

Q: Can my team share API keys?​

A: You can share keys manually (send via secure channel like password manager or encrypted message), but each user must enter the key in their own settings. Orion doesn't sync keys across team members (by design—security best practice).

Team options:

  1. Individual keys: Each team member uses their own API key (most secure)
  2. Shared key: Team shares one key, each person enters it in their settings (cost-sharing, but less secure)

Q: How do I clear conversation history?​

A: Two types of clear:

1. Visual clear only (UI):

  • Click Clear Window button in the Orion chat window bottom bar
  • Clears all messages from the UI
  • For xAI, keeps the server-side conversation chain intact (history still counts in future queries)

2. Full clear (including server):

  • Settings > AI Configuration > "Clear server history" button
  • Deletes conversation history from xAI servers
  • Next query starts completely fresh

When to use each:

  • Visual clear: Declutter UI but keep conversation context for xAI
  • Full clear: Starting a new project/topic, or if conversation chain expired

Q: What's the difference between "Graph Flow" and "Full Graph"?​

A:

Graph Flow:

  • Analyzes execution-flow nodes only (events and nodes connected via exec pins)
  • Faster processing
  • Focuses on logic flow ("what happens when this event fires?")
  • Lower token cost (smaller payload)
  • Use when: Tracing execution, debugging event flow, finding logic errors

Full Graph:

  • Analyzes all connected nodes (execution + data-linked variables and pure functions)
  • Slower processing (more data)
  • Comprehensive view (includes variable reads/writes, data flow)
  • Higher token cost (larger payload)
  • Use when: Looking for a complete Blueprint audit, finding unused variables, understanding data flow of a complex graph

Example:

  • "Why doesn't this event fire?" → Graph Flow (execution focus)
  • "What variables are used and how?" → Full Graph (data flow focus)

Q: Can I customize the prompts Orion sends?​

A: Partially. You can customize:

System messages per mode:

  • Settings > Interface > "Deep Scan Mode Context" (add custom instructions)
  • Settings > Interface > "Orbital (Ask) Mode Context" (add custom focus)

Blueprint context messages:

  • Settings > Blueprint Context > "Context Send API Prompt" (edit template)

Full template system: Planned for future releases (UDataAsset-based prompt templates with dropdown selection).

Q: Does Orion work with UE 5.6 or earlier?​

A: No. Orion is built specifically for Unreal Engine 5.7, aligning with the release of the Epic Developer Assistant for offline mode.