Tool directory · Gaming

Inworld

Last checked

Real-time voice and speech API for AI characters and NPCs.

Inworld overview

What it is

Inworld is a voice and LLM platform for developers. It offers text-to-speech (TTS-2 and TTS-2 Flash), speech-to-text, speech-to-speech and LLM inference, all paid from one credit balance. Its new Inworld Inference is, in Inworld's words, "an LLM inference API for consumer apps," serving open models such as Gemma 4, GLM, DeepSeek and MiMo.

Key features

  • An OpenAI-compatible API with streaming support.
  • Four service tiers you choose per request, API key or workspace: Flex (lowest price, fails fast and unbilled when busy), Default (no rate limits), Priority (uptime commitment) and Fast (realtime decoding).
  • A router for fallback models, conditional rules and weighted A/B splits.
  • Dedicated deployments where you can bring your own weights.
  • Zero data retention on hosted models, with LLM inference running in the US.

Getting started

  1. Point the OpenAI SDK at https://api.inworld.ai/v1, add your Inworld API key and pick a model ID such as inworld/models/gemma-4-26b-a4b-it. You can also try any model in the browser playground without an API key.

Pricing

Inworld's pricing starts with On-Demand at $0, pay as you go, which includes up to 70 minutes of TTS or 400 minutes of STT free. Paid plans are Creator at $25 a month, Builder at $100, Developer at $300 and Growth at $1,500, each granting the same amount in monthly credits. Enterprise is custom.

Inworld plans
PlanPriceDetail
On-Demand$0pay as you go, which includes up to 70 minutes of TTS or 400 minutes of STT free
Creator$25 a month
Builder$100
Developer$300
Growth$1,500
Enterprisecustom

TTS-2 runs from $25 per 1M characters on On-Demand down to $12.50 on Growth. LLM inference starts at $0.07 per 1M input tokens; Gemma 4 26B costs $0.07 input and $0.34 output per 1M tokens at Default rates. Yearly billing gives two months free.

Good to know

  • Monthly credits roll over for up to three months while you're on a paid plan, and downgrading forfeits remaining credits.
  • Concurrent requests range from 5 on On-Demand to 500 on Growth.
  • Every plan includes a commercial license, plus GDPR and SOC 2 Type II compliance. EU and India data residency are Enterprise only.

Drafted with AI assistance. No sponsorship.


Our Take

Inworld is for developers building voice-driven apps or games who want speech and open-model LLM inference on one bill, behind an API that already works with the OpenAI SDK. The free On-Demand tier and browser playground cover early experiments, and a paid plan makes sense once you need more concurrent requests or lower TTS rates, which drop at each tier from Creator at $25 a month up to Growth. Don't size up too early, though: credits roll over for only three months on a paid plan and are forfeited if you downgrade. Source