Runtime AI Chatbot Integrator (+TTS)

Bringing Live AI Chatbots and TTS into Unreal Engine

The Runtime AI Chatbot Integrator (+TTS) connects Unreal Engine projects to a range of large language model providers and text-to-speech services through Blueprint nodes, requiring no C++ coding. The plugin exposes direct API access to chatbot providers including Ollama, OpenAI, Anthropic's Claude, Google's Gemini, xAI's Grok, and DeepSeek. On the speech side, it reaches OpenAI TTS, ElevenLabs, Google Cloud, and Azure for synthesized voice output. All of this is wired through simple Blueprint nodes intended for rapid integration.

Creators working on interactive NPCs, in-game assistants, educational tutoring applications, or virtual companions can use the plugin to drive text-based conversations and spoken responses at runtime. The system is also positioned for AI-driven gameplay mechanics with voice feedback and on-the-fly narration, making it suitable for projects where character interaction and dynamic spoken content are central.

Blueprint Implementation and Authentication

Integration centers on Blueprint readiness. The plugin provides nodes for each stage of the chat and TTS pipeline, letting developers wire requests and responses visually without touching C++. Authentication across all supported services is handled through API key management within the plugin interface.

For text-to-speech examples shown in the plugin's documentation materials, the Runtime Audio Importer plugin was used to handle audio. Installing that additional plugin is necessary to follow those specific TTS examples closely. Developers can alternatively implement their own custom audio handling solution. Text-to-text chatbot functionality does not require the Runtime Audio Importer at all.

Response Handling: Complete Returns and Chunk-by-Chunk Streaming

The integrator offers two approaches for receiving chatbot responses. Developers can request complete responses — waiting for the full text to generate before delivery — or enable real-time streaming that delivers text chunk by chunk as it is produced. The streaming path is relevant for interactive applications where perceived latency matters, letting UI or gameplay systems react to partial responses as they arrive.

This flexible response handling extends into the TTS layer. OpenAI and ElevenLabs both support streaming TTS, where audio chunks arrive as they are generated. The plugin forwards those chunks for immediate playback while synthesis of the remaining text continues in the background, reducing the gap between a user action and audible feedback.

ElevenLabs Dynamic Text Streaming and Voice Discovery

ElevenLabs receives specific attention through a chunked streaming mode that allows text to be appended incrementally during an active synthesis session. This dynamic text streaming is intended for scenarios where a chatbot response is still being generated by a language model but should begin being spoken already. The approach suits real-time AI chat response synthesis where audio output needs to keep pace with an ongoing text stream rather than waiting for the full model response to complete.

For projects that need to select among available synthetic voices at runtime, the plugin includes voice discovery for Google Cloud and Azure TTS services. This allows listing and browsing voices exposed by those providers, filtering across multiple languages and locales. Google Cloud voices include Neural2, Studio, Wavenet, and Standard categories, with custom voice support. Azure provides neural voices across multiple languages, including voice styles, emotions, and custom voice options.

Cross-Platform Deployment and Demo Resources

The plugin runs on all platforms supported by Unreal Engine. That span includes Windows, Mac, Linux, Android, iOS, Meta Quest, and additional supported targets. This allows a project built with the integrator to ship across desktop, mobile, and VR from the same plugin setup.

A packaged demo project for Windows is available alongside demo source files for Unreal Engine 5.6 and above. The demo source requires both the Runtime AI Chatbot Integrator plugin and the Runtime Audio Importer plugin to be installed to function as intended. The Runtime Audio Importer dependency applies specifically to following the text-to-speech examples provided; text-only chatbot workflows do not share that requirement.

Supported Language Models and TTS Voice Catalogs

The plugin covers a broad catalog of current-generation models from each provider. OpenAI access spans GPT-5.6 variants (Sol, Terra, Luna), GPT-5.5, GPT-5.4, GPT-5.4 Mini and Nano, GPT-5.3 through GPT-5, GPT-5 Mini and Nano, GPT-4-1 and its Mini and Nano variants, GPT-4o and GPT-4o Mini, GPT-4 and Turbo and 32k variants, GPT-3.5 Turbo and Turbo 16k, ChatGPT-4o Latest, and reasoning models O1, O1 Pro, O3, O3 Mini, and O4 Mini.

Claude access includes version 5 Opus, Sonnet, and Fable; 4.8 Opus, 4.7 Opus, 4.6 Opus and Sonnet, 4.5 Haiku, Sonnet, and Opus, 4.1 Opus, 4 Sonnet, 4 Opus, 4.0 Opus, 3.7 Sonnet, 3.5 Haiku, and 3 Opus. DeepSeek coverage includes V4 Pro and V4 Flash. Gemini support spans 3.1 Pro, 3.1 Flash Light, 3.5 and 3.6 Flash, 2.5 Pro, 2.5 Flash, 2.5 Flash Lite, 2.0 Flash, 2.0 Flash Lite, Flash Latest, and Flash Lite Latest. Grok integration includes Grok 4.3, Grok 4.1 Fast Reasoning and Non-Reasoning, Grok 4 Fast Reasoning and Non-Reasoning, Grok Code Fast 1, Grok 4 0709, Grok 3, Grok 3 Mini, and Grok 2 Vision 1212. All Ollama models are supported natively, including llama3.2, llama3.1, deepseek-r1, gemma3, mistral, qwen3, and others accessed through the Ollama API.

For text-to-speech, OpenAI voices available through the plugin include Alloy, Ash, Ballad, Coral, Echo, Fable, Onyx, Nova, Sage, Shimmer, and Verse. ElevenLabs voices span Eleven V3, Eleven TTV V3, Eleven Multilingual V2, Eleven Turbo V2 and V2.5, and Eleven Flash V2 and V2.5. Google Cloud TTS offers Neural2, Studio, Wavenet, and Standard voice types with multi-language locale support and custom voice capability. Azure TTS contributes neural voices across languages with voice styles, emotional delivery, and custom voice support.

Practical Production Notes

Using the plugin requires active accounts with each service a project intends to reach. That includes OpenAI, Anthropic, Google, xAI, DeepSeek, ElevenLabs, Google Cloud, or Azure, depending on which models and TTS providers are wired into a given project. API usage with those providers may involve separate usage costs billed by the services themselves, independent of the plugin integration.

The package is set up to handle the core pipeline of sending text prompts to a selected language model, receiving either complete or streamed text responses, passing that text to a chosen TTS provider, and routing synthesized audio back through Blueprint for playback — across the full range of platforms Unreal Engine supports, from desktop builds to mobile and Meta Quest VR targets.