Trying to figure out the pricing for a powerful AI tool can feel like a wild goose chase. You're mapping out a new project, maybe dreaming up some incredibly immersive game characters or dynamic virtual agents, and you’re ready to talk numbers. You head to the company's website to check their pricing, and... you hit a wall. In the case of Inworld AI, that wall used to be a literal 404 error page. It no longer is: the official pricing page is live and lists real numbers.
The old version of this problem was a surprisingly common frustration: digging through old articles and third-party reviews to piece together a budget. The good news is you no longer have to.
This guide lays out the current Inworld AI pricing from the vendor's own page, checked on 2026-10-06: the six plans, the per-unit rates for text-to-speech (TTS), speech-to-text (STT) and LLMs, and what will actually drive your bill. I'll also help you figure out whether the structure fits your project.
What is Inworld AI?
So, what’s all the buzz about? Today, Inworld AI describes itself as a research lab and inference provider for realtime AI at consumer scale. Its products are text-to-speech, a Realtime API, speech-to-text and LLM routing. It began as a platform for AI-powered game characters, and games and media are still one of the industries it lists, alongside companions, education, health and agentic workforce use cases.
The voice layer is the center of the product now. Inworld says its Realtime TTS-2 responds in under 100ms to first audio, and on 2026-09-30 it announced the acquisition of Ultravox, a platform for building realtime voice agents. That pushes the company further toward speech-to-speech and voice agents.
The result is voice that can carry a game character, a support line or a companion app. That breadth is why the pricing is built around usage credits rather than per-character plans.
A breakdown of the Inworld AI pricing models
Alright, here’s the actual price list. Inworld publishes it on its own site, and the numbers below come from that page as checked on 2026-10-06. Prices change, so confirm on the page before you commit.
The official pricing page
Earlier versions of this guide said the page returned a 404 and sent readers to third-party directories. That is no longer true, so I've dropped those sources. The pricing page now shows every plan, a monthly or yearly toggle (the yearly option gives two months free), and per-unit rates for each model.
Inworld AI pricing: Subscription plans
Inworld sells monthly plans that each come with a credit allowance, spent across TTS, STT and LLMs. Higher tiers unlock volume discounts, more custom voices and higher limits.
| Plan | Price | Credits per month | TTS-2 per 1M chars | STT 1 per hour | Custom voices | Notable extras |
|---|---|---|---|---|---|---|
| On-Demand | Start free | None listed (list rates) | $25 | $0.15 | 100 | Up to 70 min of TTS included, 220+ LLMs via Router, community support |
| Creator | $25/mo | $25 | $20 | $0.10 | 500 | Up to 33% off TTS and STT, workspace sharing |
| Builder | $100/mo | $100 | $17.50 | $0.10 | 3,000 | Up to 40% off, higher concurrency |
| Developer | $300/mo | $300 | $15 | $0.10 | 10,000 | Up to 47% off, priority email support, professional cloning add-on |
| Growth | $1,500/mo | $1,500 | $12.50 | $0.10 | 30,000 | Up to 53% off, ZDR, HIPAA and BAA as add-ons |
| Enterprise | Custom | Custom | As low as $5 | Custom | Custom | SLA, DPA, EU and India data residency, dedicated account manager |
This tiered system gives you a path where you start free or cheap, and as your app takes off you move up a plan for a lower per-unit rate. Only the top tier is a custom negotiation.
API and usage-based pricing
Beyond the plans, the free On-Demand tier is pay as you go at list rates: TTS-2 at $25 per 1M characters, TTS-2 Flash at $15 per 1M characters, and STT 1 at $0.15 per hour. LLM usage is billed "at cost" on every plan, and Inworld says its Router adds no markup. An older third-party report of $0.002 per API interaction no longer matches anything on the pricing page.
On the surface, this model sounds great, you only pay for what you use. The catch is that the bill tracks your volume. If your app suddenly goes viral or a marketing campaign lands, your audio usage and costs rise with it, which is why the discounted plan rates matter as you grow.
The pricing for TTS
Inworld's Text-to-Speech (TTS) service is the product the pricing is built around. It includes voice cloning and voice design.
The current TTS rates are simple:
-
$25 per million characters for TTS-2, the flagship model, on the On-Demand plan.
-
$15 per million characters for TTS-2 Flash, the lower-latency model, on the On-Demand plan.
Earlier versions of this guide quoted $5 and $10 per million characters. Those figures are no longer on the current page, except that Enterprise quotes for TTS-2 can reach as low as $5.
Key factors influencing the total price
Just knowing the plans isn’t enough. Your final bill will depend on a few key factors, and some of them aren't immediately obvious when you're just starting out.
How usage volume affects pricing
The biggest cost driver is, without a doubt, usage. TTS is metered in characters, so every line your app speaks adds to the total. For a small indie game with a few dozen players, this might not be a big deal. But what happens if you succeed?
Here is the same 10 million characters of TTS-2 audio at each tier's published rate (10 x the per-1M price, my arithmetic from the pricing page):
| Plan | TTS-2 rate per 1M chars | 10M characters of TTS-2 |
|---|---|---|
| On-Demand | $25 | $250 |
| Creator | $20 | $200 |
| Builder | $17.50 | $175 |
| Developer | $15 | $150 |
| Growth | $12.50 | $125 |
The monthly plan price is separate from this usage figure, so a team at this volume should compare the discount against what the plan itself costs.
This is where a different philosophy on pricing comes in handy, especially for business applications. A tool like eesel AI, for example, is built for stability. Pricing is one fixed monthly price sized to volume, starting at $299/month for 500 credits (1 credit per ticket or chat), with no per-seat fee and a free plan of 100 credits. You can scale your support operations and know the monthly cost in advance.
Custom licenses and co-development
For large deployments, Inworld offers an Enterprise tier with custom pricing, limits and terms. It lists price match, an SLA and DPA, EU and India data residency, and a dedicated account manager and Slack channel. Enterprise TTS-2 pricing is quoted "as low as $5" per 1M characters, so the final figure depends on your volume and negotiation.
Implementation and developer resources
The price tag on an AI service is never the a full story. You also have to account for the "hidden cost" of getting it up and running. Integrating Inworld means your developers will be spending time wiring up the Realtime API and choosing models through the Router. Any custom work or complex setup falls on your team. That’s a real cost in both time and salary.
Is Inworld AI's pricing right for you?
Inworld AI’s metered, volume-driven pricing can be perfect for some projects and a poor fit for others. Let's break down when it works and when you might want to look elsewhere.
When usage-based pricing makes sense
A free pay-as-you-go tier or a cheap starter plan is a great fit for:
-
Indie developers and hobbyists: It gives you access to amazing AI tools without needing a big upfront investment. You can build your dream project on a shoestring budget.
-
Prototyping and R&D: If you're just experimenting with AI characters to see what's possible, you don't want to be locked into a big contract.
-
Projects with low, predictable traffic: For an experience with a small, stable user base, your usage costs will likely stay low and manageable.
The major downside, as we've seen, is that costs rise with your audio volume if you start to scale. For any business that needs to operate within a set budget, that takes some forecasting.
The case for predictable, flat-rate pricing
For most businesses, especially when AI is used for something critical like customer support or internal helpdesks, budget predictability is everything. A finance department needs to know what the bills will look like next quarter. A platform built on a transparent, tiered pricing model offers that peace of mind.
Let's look at the alternative approach from a solution like eesel AI, which is designed for support automation.
| Feature | Inworld AI (as reported) | eesel AI |
|---|---|---|
| Pricing Model | Monthly credit plans with metered TTS, STT and LLM rates, or custom Enterprise. | One fixed monthly price by credit batch, from $299/month for 500 credits. |
| Cost Predictability | Medium. Credits are fixed per plan, but spend follows characters and audio hours. | High. A fixed monthly price per credit batch, with optional capped overage. |
| Core Value | Voice models and realtime inference, with lower rates as you move up a plan. | A fixed price, so you can scale your support without scaling your costs. |
| Setup | Requires developer time to integrate with SDKs. | Designed to be self-serve with one-click integrations for helpdesks. You can go live in minutes. |
| Ideal Use Case | Creative and entertainment projects where usage is hard to predict. | Business-critical functions like customer service, where budget control is essential. |
With eesel AI, you get a whole suite of tools, including an AI Agent, AI Copilot, and AI Chatbot, every integration and unlimited seats included in the plan price. You can even simulate its performance on your past support tickets before you launch, so you know exactly what your ROI will be. It’s an approach built for businesses that need powerful AI without the financial headaches.
Balancing innovation with pricing predictability
Inworld AI is offering something exciting for teams building realtime voice experiences. Its pricing is now public and easy to read: six plans, published per-unit rates and a clear discount ladder. Budgeting at scale still means estimating your characters and audio hours, because the bill follows usage. It works well for experimentation and smaller creative projects, and for established businesses it pays to model your volume against each tier first.
For teams that run on financial clarity, transparent features, and a fast, self-serve setup, it’s worth looking at alternatives built with business needs in mind. Platforms that offer fixed monthly pricing and a clear return on investment provide the stability you need to confidently bring AI into core operations like customer support.
If you’re looking for a powerful AI support solution that you can set up in minutes with transparent, predictable pricing, why not try eesel AI for free?
Frequently asked questions
Where can I find the official Inworld AI pricing page?
It is live at inworld.ai/pricing and lists six plans with per-rate prices. Older write-ups that describe it as a 404 page, including earlier versions of this guide, are out of date.
What are the main components that make up Inworld AI pricing?
Inworld sells monthly credit plans: On-Demand (start free), Creator ($25/mo), Builder ($100/mo), Developer ($300/mo), Growth ($1,500/mo) and a custom Enterprise tier. Credits cover text-to-speech, speech-to-text and LLM usage, and higher tiers get lower per-unit rates.
How does usage volume impact Inworld AI pricing?
Usage is the main cost driver. TTS is billed per million characters and STT per hour, so a busy voice app spends in proportion to the audio it generates and transcribes. A higher plan lowers the rate, for example TTS-2 falls from $25 to $12.50 per 1M characters on Growth.
How much does Inworld AI text-to-speech (TTS) cost?
TTS-2 is $25 per million characters and TTS-2 Flash is $15 per million on the On-Demand plan. Paid plans discount both, down to $12.50 and $7 on Growth, and Enterprise quotes go as low as $5 per million for TTS-2.
For what kind of projects is the reported Inworld AI pricing structure most suitable?
The free On-Demand plan and the $25 Creator plan suit indie developers, hobbyists, and R&D projects with low or unpredictable traffic, since you can experiment without a large upfront investment.
What potential challenges might a business face with Inworld AI pricing if they need budget predictability?
Costs scale with characters spoken and hours transcribed, so a product that takes off will spend more. Plans give you a fixed monthly credit amount, but forecasting still means estimating your audio volume.