Inworld AI has cut prices by more than 50 percent across most of its voice AI platform, a move aimed at reducing infrastructure costs for consumer AI startups. The reductions cover text-to-speech, speech-to-text, open-source model hosting, and compute as developers face rising inference expenses. The pricing change comes as consumer AI companies try to scale apps built around subscriptions, education, companionship, health, and language learning without letting infrastructure costs consume margins.
Pricing changes span text-to-speech, speech-to-text, open-source model hosting, and dedicated compute. Inworld priced its premium real-time text-to-speech engine at roughly $10 per million characters, while streaming speech-to-text drops to approximately $0.15 per hour. Dedicated GPUs now start at $5 per hour, undercutting the roughly $11 hyperscaler rate.
Market research indicates that incumbent infrastructure remains anchored in corporate software budgets. Data from Sacra and Menlo Ventures shows that roughly 80 percent of leading AI lab revenue originates from enterprise clients. Inworld designed its custom inference stack to lower the cost per token, allowing it to sustain internal margins.
Early migrations show the financial impact. Social AI app Wishroll reported cutting its baseline AI infrastructure costs by 95 percent after transitioning to the platform. Language tutoring service TalkPal reduced expenditures by 40 percent, while Bible Chat cut text-to-speech costs by approximately 85 percent.
Inworld CEO Kylan Gibbs stated, "Cost is the first thing standing in these developers' way, so it is the first thing we are taking down." The platform has raised more than $117 million from venture firms including Lightspeed Venture Partners, Founders Fund, Kleiner Perkins, Intel Capital, Section 32, CRV, Stanford University, Microsoft’s M12 fund, and First Spark Ventures.
