Overview
The 2026 realism leader, ranked #1 on the Artificial Analysis Speech Arena. Built for real-time characters and agents, with sub-200ms latency, instant cloning from seconds of audio, 100+ languages, and steerable delivery, all at a fraction of premium pricing. Developer-first, not a studio.
Features
Realtime TTS model
Top-ranked engine for realism and expressiveness
Sub-200ms latency
First-chunk latency low enough for natural real-time conversation
Instant voice cloning
Create a custom voice from just 5-15 seconds of audio, free for all users
100+ language localization
One cloned voice speaks 100+ languages with no accent carryover
Voice Design
Design voice with accent, age, tone, and energy, no sample needed.
Steerable delivery
Natural-language prompts steer emotion and style at generation time
Unified Realtime API
TTS, STT, and LLM routing combined in a single streaming session
Pros and Cons
- Leads current realism/expressiveness benchmarks (#1 on Artificial Analysis)
- Dramatically cheaper than premium rivals (~$10/1M chars)
- Instant voice cloning, free, from 5-15 seconds of audio
- Strong for real-time, multilingual, and character-driven use
- Builder/developer-oriented, not a no-code studio
- Less suited to casual one-off voiceover tasks
- Usage-based pricing can be unpredictable at scale
- Newer to the mainstream creator conversation than ElevenLabs
Pricing Plans
Creator (Monthly)
$ 25
per month
Builder (Monthly)
$ 100
per month
Developer
$ 300
per month
Growth
$ 1500
per month
Enterprise
Custom
Contact for Pricing
Customers
- Game studios voicing dynamic, real-time NPC dialogue
- Developers building interactive AI characters and companions
- Product teams needing the most natural emotional nuance available
- Creators producing long-form narration
- Enterprises deploying voice at scale where realism is the priority
Company Information
Inworld AI
2021
Mountain View, California, USA
Frequently Asked Questions
How does Inworld’s pricing actually work?
It’s credit-based. Each paid plan grants monthly credits equal to the plan price (e.g., the $300 Developer plan gives $300 in credits), which are consumed at your tier’s per-usage rates across TTS, STT, and LLMs. Higher tiers unlock lower per-character rates.
What does TTS cost per character?
On the free On-Demand tier, Realtime TTS-2 is $25 per 1M characters and TTS 1.5 Mini is $15 per 1M. Rates drop as you move up tiers, reaching as low as $5 per 1M characters (TTS-2) on Enterprise.
Is there a free tier?
Yes. On-Demand is free and includes up to 70 minutes of TTS (or 400 minutes of STT), 100 custom voices, voice cloning, and commercial license.
Do unused credits roll over?
Yes, for up to 3 months, provided you keep an active paid subscription of equal or higher value. Each monthly grant has its own 3-month expiry window.
How fast is voice cloning and how many languages does a cloned voice support?
Instant cloning takes a 5–15 second sample and is ready in seconds. A single cloned voice can then speak 100+ languages (with TTS-2) with no accent carryover.
Is commercial use included?
Yes, a commercial license is included on all tiers, starting from the free On-Demand plan.
Does Inworld offer HIPAA/compliance options?
GDPR and SOC 2 Type II apply across tiers; HIPAA & BAA and zero-data-retention are available as add-ons on Growth, and fully on Enterprise (which also adds SLA/DPA, on-prem, and EU/India data residency).
What happens if I go over my credits?
You can buy additional credits or enable auto-reload. Extra credits are billed at your current tier’s rates and expire one year after purchase.
Strength
Text-to-Speech Realism
rating
⭐️ 4.5
Starts At
$ 25
per month








