How the arithmetic works
10 of the 10 carry enough in their price matrix to work the shape through end to end, and the table below does exactly that with each model's own figures. No two billing shapes on this site share a model — a model has exactly one — so nothing here overlaps another billing page.
credits = a per-1,000-character rate x the length of the text you send. This is the one shape where the bill is set entirely by your input and not at all by the output.
| Model | Rate | A 2,000-character script | A 10,000-character script |
|---|---|---|---|
| Qwen 3 TTS 0.6B | 12 credits per 1,000 characters | 24 credits | 120 credits |
| Inworld TTS | 12 credits per 1,000 characters | 24 credits | 120 credits |
| Qwen 3 TTS Voice Design | 12 credits per 1,000 characters | 24 credits | 120 credits |
| Gemini 3.1 Flash TTS | 12 credits per 1,000 characters | 24 credits | 120 credits |
| Grok TTS | 12 credits per 1,000 characters | 24 credits | 120 credits |
| Inworld TTS 1.5 Max | 12 credits per 1,000 characters | 24 credits | 120 credits |
| Seed Audio 1.0 | 12 credits per 1,000 characters | 24 credits | 120 credits |
| Cartesia Sonic 3.5 | 12 credits per 1,000 characters | 24 credits | 120 credits |
| Inworld TTS 2 | 12 credits per 1,000 characters | 24 credits | 120 credits |
| ElevenLabs Multilingual | 12 credits per 1,000 characters | 24 credits | 120 credits |
Worked from each model's own price matrix. A cell reading “—” is one where the arithmetic produces a figure outside the complete-job range the same matrix declares, so nothing is printed.
All 10 models
Cheapest complete job first. Every row links to that model's spec page — tier and mode variants share a page with the model they are a variant of, so this is products rather than SKUs.
| Model | Provider | Credits per job | Its priced options |
|---|---|---|---|
| Qwen 3 TTS 0.6B | Qwen | 2 credits (headline rate) | per 1,000 characters 12 |
| Inworld TTS | Inworld | 3 credits (headline rate) | per 1,000 characters 12 |
| Qwen 3 TTS Voice Design | Qwen | 3 credits (headline rate) | per 1,000 characters 12 |
| Gemini 3.1 Flash TTS | 4 credits (headline rate) | per 1,000 characters 12 | |
| Grok TTS | Grok | 4 credits (headline rate) | per 1,000 characters 12 |
| Inworld TTS 1.5 Max | Inworld | 4 credits (headline rate) | per 1,000 characters 12 |
| Seed Audio 1.0 | ByteDance | 4 credits (headline rate) | per 1,000 characters 12 |
| Cartesia Sonic 3.5 | Cartesia | 5 credits (headline rate) | per 1,000 characters 12 |
| Inworld TTS 2 | Inworld | 5 credits (headline rate) | per 1,000 characters 12 |
| ElevenLabs Multilingual | KIE | 6 credits (headline rate) | per 1,000 characters 12 |
How these models are billed
Every model on this page uses the same billing shape.
- Charged per 1,000 characters of input text10 models
All figures are Versely credits. A credit figure marked as a headline rate is not the cost of a complete job. Your plan's credit allowance is on the pricing page.
More billing shape pages
Every model on Versely · Model providers · Ranked buyer guides
Frequently asked questions
Which AI models are billed per 1,000 characters?+
10 on Versely — all of them audio & voice models, from 7 providers. credits = a per-1,000-character rate x the length of the text you send. This is the one shape where the bill is set entirely by your input and not at all by the output.
Is a per 1,000 characters model cheaper?+
Not in itself — the shape is how you are charged, not how much. Across these 10 models a complete job runs 0 credits. What the shape does tell you is which lever moves your bill, and on this page that lever is the same for every model listed.
What is the cheapest model billed per 1,000 characters?+
Qwen 3 TTS 0.6B has the lowest complete-job cost of the 10 at 2 credits (headline rate). The worked example table above shows the same arithmetic applied to every model that publishes enough of its matrix to support it.
Can I compare a per 1,000 characters model against one billed another way?+
Only on the cost of a complete job, never on the headline figure. That is the whole reason these pages are split by shape — a rate compared against a job total is two different quantities in one column. 10 of the models here publish no complete-job cost at all, only a rate, so for those there is no like-for-like figure to compare. Versely's ranked buyer guides sort on complete-job cost for exactly this reason.
Run every one of them on one subscription
The all-in-one AI studio for creators. 60+ models for video, image, voice, music and lipsync in a single app.