AI research lab Inworld AI drops pricing more than 50% for most developers across text-to-speech, speech-to-text, the LLM, and compute, the opening move in becoming the company every consumer AI team builds on

As model costs climb, the entire consumer AI market is stalling. The teams building for health, education, companionship, and language learning are being priced out of their own category. While enterprise companies absorb rising inference costs with padded budgets and per-seat pricing, the consumer developers that acquire users through paid ads, monetize on $5-10 subscriptions, and carry AI spend as their single largest cost line have nowhere left to go.

Small teams prove out ideas, start to scale, hit a cost wall, and either stop growing or shut down. The apps that would bring AI into everyday life are the ones that cannot afford to exist at scale.

Today Inworld AI cut prices more than 50% for most developers across its voice AI platform, effective immediately. It is the first public move in Inworld's larger commitment to giving consumer developers the technology, economics, and support infrastructure to scale.

"Every consumer team that comes to us has the same story," said Kylan Gibbs, CEO and co-founder of Inworld AI. "They built something people use. The technology works. And the cost makes it impossible to grow. Somebody has to take a step. We decided it should be us, because we built the research that makes it possible to do it without compromise."

Inworld's voice models rank among the top on the independent Artificial Analysis Speech Arena because the underlying research optimized for both quality and serving efficiency at consumer scale. The company designed its inference stack to reduce cost per token at scale, which means it can lower prices for developers while sustaining its own margins. This is an engineered advantage that Inworld's customers benefit from.

Consumer developers are already Inworld's fastest-converting segment. They are choosing Inworld because the alternatives were built for enterprise support, with enterprise pricing and enterprise assumptions about session length, expressiveness, and use case. Roughly 80% of a leading lab's revenue comes from business and developer customers, and only about 3% of consumer AI users ever pay (Sacra, 2026; Menlo Ventures, 2025). A consumer voice session runs longer, carries more emotional range, and requires different economics than a support call. The incumbents retrofit. Inworld built for it.

Wishroll, the team behind the social AI app Status, is a five-person company. They reached more than 500,000 daily active users, with people spending over 90 minutes a day in the app. After switching to Inworld, Wishroll cut AI costs by 95 percent. "Working with Inworld has been the most positive experience I've had with another company," said Fai Nur, CEO of Wishroll. TalkPal, an AI language tutor serving more than five million learners, cut TTS costs by 40 percent after moving to Inworld and increased voice usage across its platform. Bible Chat serves millions of users across the full stack and cut text-to-speech costs about 85 percent. (Customer figures are self-reported.)

"Cost is the first thing standing in these developers' way, so it is the first thing we are taking down," Gibbs said. "But a consumer app does not succeed on cheap inference alone. These teams need help with growth and distribution. We are building toward a model where a small team can come to Inworld and get the technology, the economics, and the support to build a real consumer business."

What Changes

New pricing is live today, more than 50% lower for most developers across every layer, and lower still as usage grows.

• Text-to-speech. Realtime TTS-2, top-ranked on the Artificial Analysis Speech Arena, starts at about $10 per million characters, and TTS-1.5 Mini at about $5, less than half the published rate of comparable premium voices.
• Speech-to-text. From about $0.15 per hour, among the lowest streaming rates available, with voice profiling that reads who is speaking and how.
• Realtime inference for LLMs, a new first-party product, hosts and optimizes the top open models consumer apps run in production at roughly 50% below the public third-party rate.
• Realtime Router: Hundreds of routed third-party models stay available at no markup. Three of the ten largest token-consuming consumer applications are already moving their LLM workloads onto it, including one that processes more than 600 billion tokens a day.
• Compute. Dedicated GPUs from $5 per GPU-hour, against roughly $11 for an on-demand hyperscaler equivalent. Teams transition from per-token billing when the economics make sense for them.

Volume pricing tiers reduce cost further as developers scale, spend on one layer lowers the others on one combined bill, and committed usage unlocks additional reductions. The structure is designed so that economics improve as an app grows.

Beyond pricing, Inworld is building partnerships and programs to support consumer developers in areas where small teams struggle most: user acquisition, growth, analytics, and business infrastructure. The company plans to announce specific partnerships in the coming months.

Inworld AI has raised more than $117 million at a $500 million valuation from Lightspeed Venture Partners, Founders Fund, Kleiner Perkins, Intel Capital, Section 32, CRV, Stanford University, Microsoft's M12 fund, and First Spark Ventures (Eric Schmidt).

Availability

New pricing is live today at inworld.ai/pricing. Developers can get started free at inworld.ai.

About Inworld AI

Inworld AI is a research lab building the best real-time voice AI models and APIs for developers of consumer-facing applications. Founded by former Google DeepMind and Dialogflow engineers, including CEO Kylan Gibbs and Executive Chairman Ilya Gelfenbeyn (who previously built and sold API.AI to Google), Inworld builds voice models and infrastructure that give small teams frontier quality and the economics to serve users at scale. Its top-ranked TTS, STT with voice profiling, LLM Router, realtime inference, and Realtime API power consumer apps used by millions of people every day.

View source version on businesswire.com: https://www.businesswire.com/news/home/20260610968386/en/

AI costs are climbing and the consumer AI economy is stalling because of it. Enterprise infrastructure is being repurposed for consumer developers at enterprise pricing, making consumer scale impossible. Today Inworld AI is taking down the cost wall.