Inworld AI
VerifiedInworld AI provides realtime voice and LLM inference infrastructure for consumer applications, including games, companions, education and wellness apps. It offers text-to-speech, speech-to-text and speech-to-speech models, along with access to hosted LLMs through its Inference platform. Developers can build NPCs and voice agents with configurable personality, memory and low-latency speech, using SDKs for Unity and Unreal plus a general API. The platform targets teams building voice-driven consumer products that need fast response times, broad language support and scalable, cost-efficient pricing. It suits game studios, companion app builders, educators and support teams who need natural-sounding, low-latency voice interaction baked into their product.
Inworld AI at a glance
- What it is
- Inworld AI provides realtime voice and LLM inference infrastructure for consumer applications, including games, companions, education and wellness apps.
- Who it's for
- Worth a look if projects involving NPC & Dialogue AI come up often in your week.
- Primary use case
- Inworld gives non-player characters a configurable personality, goals, knowledge base and emotional state, then drives their speech and behaviour at runtime.
- CategoryGame Development AI, AI Games
- PricingFree Tier
- Starting price$0 (On-Demand pay as you go)
- Credit cardNo credit card required
- Foundation modelProprietary
- Bring your own keyNot supported
- RatingNo reviews yet. Be the first to rate.
Video Spot
Overview
Inworld AI provides realtime voice and LLM inference infrastructure for consumer applications, including games, companions, education and wellness apps. It offers text-to-speech, speech-to-text and speech-to-speech models, along with access to hosted LLMs through its Inference platform. Developers can build NPCs and voice agents with configurable personality, memory and low-latency speech, using SDKs for Unity and Unreal plus a general API. The platform targets teams building voice-driven consumer products that need fast response times, broad language support and scalable, cost-efficient pricing. It suits game studios, companion app builders, educators and support teams who need natural-sounding, low-latency voice interaction baked into their product.
Inworld gives non-player characters a configurable personality, goals, knowledge base and emotional state, then drives their speech and behaviour at runtime. SDKs for Unity and Unreal handle voice, lip-sync and safety guardrails so characters stay inside the fiction.
In detail
Inworld AI builds infrastructure for voice-driven and conversational AI products, originally known for giving game NPCs personality, memory and real-time voice, and now expanded into a broader platform called Inworld Inference for consumer applications generally. It offers realtime text-to-speech, speech-to-text and speech-to-speech models, as well as hosted access to open LLMs such as Gemma, GLM and DeepSeek, plus decision and video generation models.
The platform is aimed at developers and teams building companion apps, learning and education tools, health and wellness coaches, agentic workforce tools like support and sales voice agents, and games or narrated media. SDKs for Unity and Unreal support game developers specifically, while a general API, CLI and coding-agent integrations (for tools like Claude Code and Cursor) support broader application developers.
Under the hood, Inworld's voice models focus on low latency, claiming as low as 50ms to first audio for its fastest text-to-speech model, support for roughly 500 languages and dialects, and features like lip-sync timings, pronunciation dictionaries, speaker diarization and voice cloning. Speech-to-speech works over a single WebSocket or WebRTC connection that listens, reasons and replies. LLM Inference lets developers run open models on Inworld's hosted fleet, optimized for speed and cost, with claims of lower cost and faster time-to-first-token than other providers of the same models.
What differentiates Inworld is the combination of voice and LLM infrastructure in one platform, with a credits-based pricing model that scales from free pay-as-you-go usage up to enterprise plans with compliance features like SOC 2 Type II, HIPAA/BAA add-ons and data residency options. Reliability is highlighted through stated uptime figures and regional deployment across the US, Europe, India and Asia Pacific for voice, though LLM serving is US-based only.
Common uses
- Giving game NPCs personality, memory and real-time voice in Unity or Unreal
- Building companion or social apps with natural-sounding conversational voice
- Creating tutoring or language-learning tools with interactive voice dialogue
- Building voice agents for customer support, sales or operations
- Adding narrated voice and dialogue to interactive media
- Building health and wellness coaching or meditation apps with voice interaction
Pros
- Combines voice (TTS/STT/speech-to-speech) and LLM inference in a single platform
- Low stated latency and broad language support for voice models
- Free on-demand tier available for evaluation and prototyping
- Compliance features like SOC 2 Type II and GDPR included on all plans, with HIPAA/BAA and data residency on higher tiers
- Credit rollover and volume discounts as usage scales
Cons
- LLM serving is limited to US regions, which may add latency for non-US LLM workloads
- Advanced compliance options like HIPAA/BAA, SLA and data residency require higher-tier or Enterprise plans
- Some features like multilingual voice cloning and speaker diarization are still in beta or experimental
- Pricing structure with tiered credits and per-feature rates may require careful cost estimation for complex use cases
Generated from the listing's details and not yet checked by a Zoipi editor.
Key Features
- Realtime text-to-speech: Low-latency speech generation (as low as 50ms to first audio) in close to 500 languages
- Speech-to-text: Streaming transcription with voice traits like emotion, style, age and accent, plus speaker diarization and async transcription
- Speech-to-speech: Single connection that listens, reasons and talks back using any LLM per session
- LLM Inference: Hosted open models (Gemma, GLM, DeepSeek and others) optimized for speed and cost
- Voice cloning: Instant and professional voice cloning, including multilingual cloning
- Lip-sync and pronunciation control: Lip-sync timings for character animation and custom pronunciation dictionaries
- Coding agent integration: CLI and skill for coding assistants like Claude Code and Cursor to synthesize speech, transcribe audio and clone voices
- Compliance options: SOC 2 Type II and GDPR on all plans, with HIPAA/BAA and data residency add-ons on higher tiers
Best For
- Specialists looking for a dedicated tool rather than a general assistant
- Users working with NPC & Dialogue AI
- Anyone testing the space without an upfront budget
Pricing Model
Free Tier. Generous ongoing free tier you can start with, no credit card needed. Paid plans unlock higher limits.
Pricing Breakdown
Screenshots
Live preview of Inworld AI's website.
Frequently asked questions
Community Reviews & Discussion
0 reviews · replies are moderated before they appear.
- No reviews yet — be the first to review Inworld AI.