Inworld AI
Inworld AI
Developer platform for realtime AI voices and speech, with a games and media use case
Overview
Inworld AI is a developer platform for voice and language AI. Older write-ups describe it mainly as an engine for game characters, but its own site now centres on text-to-speech, speech-to-text, speech-to-speech and an inference service for language models. Games and media appear as one of several solution areas, described as voices and live dialogue for games and narrated media, next to companions, learning, wellness and agent workflows. Anyone choosing it for a game should treat it as a voice and dialogue building block rather than a ready-made game character kit.
Pricing is usage based with monthly plans that include credits equal to the plan price. The free On-Demand tier is pay as you go, with listed rates per million characters of text-to-speech, per hour of speech-to-text and per million tokens for language models. Paid tiers (Creator, Builder and Developer) bring volume discounts, more custom voices, more professional voice clones and higher limits. According to Inworld, GDPR compliance and SOC 2 Type II apply on every plan, and unused credits roll over for three months.
It suits developers who are comfortable with an API and usage billing. Check the voice-cloning rules before you use it, because cloning needs consent from the person whose voice you copy. Its terms say California law governs. This page is a summary based on public listings and the vendor’s own description, dated 16 November 2026. It was not tested hands-on, and details may have changed.
At a glance
Key Features
• Speech-to-text
• Speech-to-speech
• LLM inference service
• Custom voices and professional voice clones
• Games and media solution area
• Playground and API reference
• Plans with monthly credits
Visual Walkthrough
Best For
• Developers building voice agents or companions
• Teams producing narrated media
• Builders comfortable with usage-based pricing
• Players looking for a game to play
• Buyers who need fixed monthly pricing
• Anyone unwilling to read voice-cloning rules
Use cases
2. Build a voice agent for support or sales.
3. Create narrated content with custom voices.
4. Transcribe speech to text through an API.
5. Run language models through its inference service.
Pricing
On-Demand: $0, pay as you go. Text-to-speech $25 per 1M characters (Flash $15), speech-to-text $0.15 an hour, language models from $0.07 per 1M tokens. Up to 70 minutes of TTS included, 100 custom voices, 2 professional voice clones.
Creator: $25 a month with $25 in credits, up to 33% off text-to-speech and speech-to-text, 500 custom voices, 4 professional voice clones.
Builder: $100 a month with $100 in credits, up to 40% off, 3,000 custom voices, 10 professional voice clones.
Developer: $300 a month, with deeper volume discounts.
Enterprise: ask Inworld.
Prices change often. Check the official pricing page for current prices.
Pros and cons
• Clear per-unit rates on the pricing page
• Commercial license listed
• Credits roll over for three months, per Inworld
• GDPR and SOC 2 Type II on every plan, per Inworld
• Usage-based billing is harder to predict
• Needs developer skills
• Voice cloning carries consent duties
• Community support only on the free tier
Integrations
• Unity and Unreal support was not confirmed on the pages read
• Open-model inference through its own service
FAQ
A: It was known for that. Its current site centres on voice and language AI, with games and media as one solution area.
Q: Is there a free plan?
A: Yes. The On-Demand tier costs $0 and bills by usage.
Q: Can I clone any voice?
A: Only with consent. Read its terms and your local rules first.
Alternatives
Convai: conversational characters aimed at game engines.
Scenario: game art and assets in a studio's own style.






