NVIDIA ACE LLM brings local large language model inference to Unreal Engine projects on Windows. Generate conversational dialogue, drive agentic NPC behaviors, and power interactive scenes with on-device language models without sending prompts to an external service for every response.

Designed for a simple Unreal Engine workflow, the plugin includes Blueprint-friendly chat sessions, streamed token output, tool calling for gameplay actions, and sample content to help you get started quickly. Use it for character dialogue, branching conversations, agentic NPC behaviors, narrative prototyping, or interactive AI-driven experiences.

Key Features:

• Local LLM inference for responsive, private text generation

• Blueprint and C++ support for fast integration

• Real-time token streaming for live conversational UI

• Support for local GGUF models with configurable context size

• Per-NPC chat history, system prompts, and sampling controls

• Tool calling for agentic NPCs that can request gameplay actions

• Project Settings catalog for shared tool definitions across NPCs

• Sample level and plugin content for quick setup

Use Cases:

• Conversational NPCs and dynamic dialogue

• In-game actions via tool calls

• AI-driven gameplay systems and tutorials

Requirements:

• Unreal Engine 5.5, 5.6, or 5.7

• Windows 10 or 11, 64-bit

• NVIDIA RTX GPU recommended for best performance

• DirectX 12 RHI required for GPU-accelerated inference

• Local .gguf chat model (not bundled)