Meta Quest AI Building Blocks vs Neocortex for VR NPCs
If you are building for Meta Quest in Unity, Meta's own tools are the obvious first stop. The Meta XR SDK now includes AI Building Blocks, and the long-standing Voice SDK handles voice commands and dictation.
They are useful plumbing. They are not characters.
What Meta offers in 2026
From Meta's developer documentation (checked October 2026):
- AI Building Blocks for LLM, speech-to-text, text-to-speech and object detection. Each block connects to a provider you choose, such as OpenAI, Ollama, Hugging Face or ElevenLabs, and small models can run on-device.
- Requirements: Unity 6, Quest 3 or 3S, and a recent Meta XR Core SDK, with breaking API changes between recent versions.
- Voice SDK built on Wit.ai for voice commands, dictation and TTS.
- A changing LLM backend. Meta retired its Llama API in July 2026, according to developer reports, and moved to a new model API.
What you still build
- The character: persona, knowledge, memory and guardrails.
- Wiring LLM, STT and TTS blocks together, and paying each provider separately.
- Voice cloning through a third party.
- Game integration for actions and scene awareness.
- Any version for other headsets, PC or Unreal.
Meta AI Building Blocks vs Neocortex
| Meta AI Building Blocks | Neocortex | |
|---|---|---|
| What it is | Provider connectors for Unity on Quest | Complete AI character platform |
| Character design | Build your own | Visual node editor |
| Voice | Connect a TTS provider | Streaming TTS and STT, voice cloning and design |
| Providers and billing | Several, billed separately | One account |
| Platforms | Quest 3 and 3S | Unity on desktop, mobile and web, plus Unreal, Godot (beta) and Web API |
| Gameplay hooks | Build your own | Actions, interactable objects, per-line emotions |
Why VR teams choose Neocortex
Characters that live in the scene. Interactables let a character notice tagged objects around the player and act on them, which matters in VR, where players point, pick up and look at things.
Talk naturally. Players speak, characters answer in their own cloned or designed voice, streamed so replies start fast.
Designers own behaviour. Personality, emotion, knowledge, intents and actions are nodes anyone on the team can edit and test in the browser.
Not tied to one headset. The same character works across platforms and engines, so a Quest project is not a dead end.
We have published a VR demo with Unity's XR Interaction Toolkit if you want to see a character in a headset.
Get started
Meta details were last checked in October 2026.
Written by
Sercan Altundas
Date
Oct 17, 2026
Read time
4 min
