Inworld AI
Realtime voice models and open LLM inference for consumer apps and developers
- Pricing
- Freemium
- Platforms
- Web, API
- Listing updated
- 1 day ago
Similar tools
About Inworld AI
Inworld AI is a freemium AI tool for chat. It is available on Web and API.
The README-free product site describes Inworld as a research lab and inference provider aimed at consumer AI products. Its voice stack covers realtime text to speech, speech to text, and a speech to speech connection that listens, reasons and replies over one WebSocket or WebRTC session. Hosted open LLMs such as Gemma, GLM and DeepSeek are served through an OpenAI-compatible Inference API on infrastructure Inworld operates in several regions.
Developers get API keys and a playground to test voices and models before writing code. Pricing is metered in dollar credits that cover TTS, STT and LLM usage, with plan tiers that reduce per-character rates and raise concurrency and voice-clone limits. The site also notes that coding agents such as Claude Code and Cursor can be given Inworld's skill to wire the APIs in.
The product suits teams building companions, learning apps, fitness or health assistants, and voice agents that need low latency and high uptime.
What it does
- Realtime text to speech
- Realtime TTS-2 and TTS-2 Flash stream audio as it is generated, with the non-Flash model steerable in plain language.
- Speech to speech connection
- One WebSocket or WebRTC connection handles listening, reasoning with a chosen LLM, and spoken replies.
- Streaming speech to text
- Realtime STT-1 returns streaming transcripts with voice traits.
- OpenAI-compatible open model inference
- Open models such as Gemma 4 and GLM 5.3 Flash are served through an OpenAI-compatible API.
- Custom and cloned voices
- Plans include custom voices and professional voice clones, with limits that rise by tier.
- Coding agent integration
- A skill lets coding agents such as Claude Code and Cursor add Inworld to a project.
Good for
- Adding spoken dialogue to companion or entertainment apps
- Building a voice agent with a choice of LLM over one connection
- Narrating story content or lessons in many languages
- Transcribing user speech in a streaming app
- Serving open LLMs behind an OpenAI-compatible endpoint
Pricing
Freemium, from $25 a month
On-Demand free tier with 70 min TTS; Creator from $25/mo in credits; Enterprise custom.
Prices change. Check inworld.ai before you buy.
Inworld AI FAQ
What is the fastest text to speech API for realtime apps?
The site presents Realtime TTS-2 Flash as its fastest voice model, citing first audio in as little as 50 ms on Coval's public benchmark. Realtime TTS-2 starts speaking in as little as 90 ms at P90 and covers about 500 languages and dialects.
How much does Inworld text to speech cost?
On-demand list prices are $25 per million characters for Realtime TTS-2 and $15 for Realtime TTS-2 Flash, falling with higher plans. Realtime STT-1 is listed at $0.10 per hour of audio on paid plans, and open LLMs start from $0.07 per million input tokens.
Which open models can I run on Inworld Inference?
The site lists GLM 5.3 Flash, Gemma 4 31B, Gemma 4 26B, MiMo 2.6 Pro and DeepSeek V4.1 Flash, served through an OpenAI-compatible API.
Is there a free tier?
The On-Demand plan includes up to 70 minutes of TTS and 400 minutes of STT for free, plus a 2,000-character TTS Playground limit.
Is Inworld AI free?
Inworld AI has a free plan, with paid plans for more usage or features. Paid plans start at $25 a month.
What are the best alternatives to Inworld AI?
The closest alternatives to Inworld AI are Eleven Labs, AiCogni, WellSaid Labs, Convai and Verbatik.
Where this comes from
- Description: written from the project's website.