What is included in Early Access?

Spectra AI is an incredibly ambitious project, and Early Access means you are getting in on the ground floor while we actively build and refine the ecosystem. Please note that the software is still under heavy development—you may encounter bugs, and not all planned features are fully implemented yet. However, the current build includes a massive suite of working tools:

Advanced AI Chat: Converse with Local Language Models completely offline, or optionally connect to the lightning-fast Groq API for free.
Modular Intelligence: Bring your own intelligence! You can load your own custom GGUF LLMs into the system.
Total Customization: Create a fully local profile and tweak system settings and AI prompts to fit your exact needs.
Full Voice Pipeline: Talk directly to Spectra using your microphone with our integrated Speech-to-Text (STT).
Neural Voice Synthesis: Spectra talks back with incredibly human-like, integrated Text-to-Speech (TTS). We also include a Narrator tool with multiple voices.
Spectra Music Core: Listen to high-quality audio while you work or play, featuring 45 genres and over 1 million songs.
Arcade Hub: Take a break with fully playable 3D arcade games featuring local high scores. Currently includes Tetro 3D , with more games in active development!

Does Spectra require an internet connection?

Spectra AI is a 100% local, offline software suite. Once downloaded from the Epic Games Store, it runs entirely on your own hardware. There are no cloud servers, no API calls, and absolutely no telemetry. Your data stays on your machine.

Do I need to pay for a subscription or an API key (like OpenAI/ChatGPT)?

Absolutely not. Because Spectra AI runs locally using advanced GGUF models and ONNX runtimes, you do not need any third-party API keys or monthly subscriptions. You buy the software once, and it is yours to use forever with zero hidden operating costs.

What are the system requirements for local LLMs?

Because Spectra runs models locally via llama.cpp, your requirements depend entirely on the model you choose. For basic functionality with an 8B model (like Llama 3 8B), you will want at least 8GB to 12GB of VRAM. For highly scalable 70B+ models, you will need high-end multi-GPU hardware or massive amounts of unified system memory.

Can I load my own custom AI models into Spectra?

Yes! Spectra's modular design allows power users to swap in their own compatible GGUF format Large Language Models, allowing you to tailor the intelligence and personality of the system to your specific needs.

Can I use Spectra for commercial projects?

Yes! However, you must adhere to the specific licenses of the underlying models you choose to run (e.g., Meta's Llama license, Mistral's license). Spectra provides the engine; you provide the brains.

Do I need specific drivers to run the AI models?

For the absolute best performance and inference speed on Nvidia GPUs, we highly recommend installing the CUDA Toolkit. Spectra runs flawlessly on CUDA 13.2, and is fully backward compatible with previous CUDA versions.

Is Spectra just a chatbot, or is it a game?

It is a fully interactive Unreal Engine 5 Game Hub! While the core of Spectra is an advanced, localized MCP AI Server with a fully animated MetaHuman companion, the hub also includes a music system and a fully playable retro arcade game called Tetro 3D.

Do I need to install Python or know how to code to run this?

No. While Spectra's backend is incredibly complex, we have bundled everything into a standalone executable for the Epic Games Store. You do not need to install Python, run command lines, or deal with complicated environments. Just click "Play" in your Epic Launcher and the suite will initialize automatically.

Do I need an Nvidia GPU? Are CUDA drivers required?

Spectra AI leverages heavy machine learning models (Local LLMs, Whisper Speech-to-Text, and XTTS-v2 Neural Voice Synthesis). For a smooth, real-time experience, a modern Nvidia GPU is highly recommended. To get the absolute best performance and utilize hardware acceleration, we strongly recommend installing the latest Nvidia CUDA Toolkit (version 13.2 or compatible).

What are the minimum system requirements?

Due to the local processing of AI models and the rendering of Unreal Engine 5 MetaHumans, we recommend a minimum of:

OS: Windows 10/11 (64-bit)
RAM: 16GB minimum (32GB strongly recommended for larger LLM contexts)
GPU: Nvidia RTX 2060 or better with at least 8GB of VRAM (12GB+ recommended)
Storage: SSD heavily recommended for fast model loading times.

Can I use the audio or content generated by Spectra commercially?

Spectra utilizes open-weights models (like Llama and Whisper). You must adhere to the standard open-source licenses of the underlying models you choose to run. Generally, yes, but please review the specific license of the LLM you load into the system for commercial restrictions.