Inworld TTS-2 is not a $300 seat
Developer is $300 of Inworld credits at $15/1M TTS-2 chars, not a studio seat. On-Demand TTS-2 is $25/1M. Flash is $15/1M.
Inworld's $300 Developer plan is $300 of credits at a cheaper TTS-2 rate. It is not a studio seat, not captions, not a talking generate. On-Demand Realtime TTS-2 is $25 per million characters. Flash is $15. Creator is $25 a month. Steering lives on the TTS-2 model, not on the $300 rung. Keep the voice file. Stop treating Developer as a room you have to rent.
Inworld's pricing page, fetched 19 September 2026, and Inworld's billing docs (the same plan table also rendered at TTS billing the same day) are the sources for every public dollar below. If a number is not on those two pages, it is not here. The job page for when to pick the row is When to use Inworld TTS-2. This page is the bill.
What the $300 actually is
Developer is labelled "Production applications" on the public ladder. The sticker is $300/mo. The grant is $300 in credits per month. That is Inworld's own FAQ language, not a translation: credits are a dollar-denominated balance used across TTS, STT, the LLM Router, and the Realtime API. You are not buying a named human in an editor. You are buying a prepaid bucket and a cheaper unit rate.
On the 19 September 2026 monthly cards, Realtime TTS-2 on Developer is $15 per million characters, marked down from the On-Demand $25. Realtime TTS-2 Flash is $8, marked down from $15. STT 1 is $0.10/hr, marked down from $0.15/hr. The card also prints up to 47% off TTS and STT rates, 10,000 custom voices, increased concurrency, professional voice cloning as an add-on (not included), and priority email. Growth is the first public tier that includes 1 professional voice clone in the list.
Do the only arithmetic the sticker allows. If the entire $300 grant is spent on realtime TTS-2 at $15 per million characters, that is 20 million characters. Inworld's compare table states it assumes one minute of audio is ~1,000 characters. Under that assumption, 20 million characters is about 20,000 minutes of TTS-2. That number collapses the moment the same wallet pays for STT hours, router tokens, or Realtime sessions, because those meters share the grant. The $300 is not "20,000 minutes of studio." It is $300 that happens to buy 20 million TTS-2 characters if nothing else in the account speaks.
The rest of the ladder, same fetch, same monthly cards:
| Plan | Sticker | TTS-2 / 1M chars | Flash / 1M chars | Credit grant |
|---|---|---|---|---|
| On-Demand | Start free | $25 | $15 | Up to 70 min TTS included |
| Creator | $25/mo | $20 | $10 | $25 in credits (up to 33% off) |
| Builder | $100/mo | $17.50 | $9 | $100 in credits (up to 40% off) |
| Developer | $300/mo | $15 | $8 | $300 in credits (up to 47% off) |
| Growth | $1,500/mo | $12.50 | $7 | $1,500 in credits (up to 53% off) |
| Enterprise | Custom | As low as $5 | Sub-$5 | Custom |
Creator at $20 per million is $25 of credits, so 1.25 million TTS-2 characters if the whole grant is realtime TTS-2. Builder at $17.50 is about 5.7 million. Growth at $12.50 is 120 million. Those are wallet sizes, not seats. Enterprise's as low as $5/1M is a custom line. It is not the public default, and it is not what a $300 card buys.
Billing docs, same day: paid plans include credits equal to the monthly plan price each cycle. Unused subscription credits on Creator and above roll over for up to 3 months while the paid plan stays active at equal or higher value. Each month's grant keeps its own 3-month window. Extra credits you buy (minimum $10) are billed at the current tier's rates and expire after 1 year. Downgrade or cancel at period end and remaining credits are forfeited. A failed subscription charge is retried for 7 days, then the account drops to On-Demand and API usage is disabled until you upgrade again. None of that is how a studio seat behaves. A seat is a person. This is a rate card with a prepaid pile.
The page also has a Yearly toggle labelled 2 months free. Annual subscriptions are billed once per year, per the billing docs. The monthly cards above are the numbers that were printed next to the plan names on the fetch. This page does not invent a yearly per-character table the public grid did not show.
On-Demand already sells TTS-2
The mistake is thinking TTS-2 starts at $300. On-Demand is the first column. It is labelled "Evaluation and prototyping" and Start free. The same TTS-2 SKU is on that column at $25/1M chars. Flash is $15/1M. The included trial is up to 70 minutes of TTS (billing docs also give On-Demand up to 400 minutes of STT for free). Commercial license, instant voice cloning, voice design, and Realtime API access are on the On-Demand card. Custom voices on that card: 100. Playground character limit on the compare table: 2,000. Concurrent generations in the billing table: 5. WebSocket connections: 50.
That is already the product. You synthesize characters, you pay per million, you download audio. You do not unlock TTS-2 by climbing to Developer. You unlock a cheaper unit rate, more custom voices, more concurrent generations, a larger playground paste, team furniture, and (on Developer) the option to add professional cloning. If the job this week is one narrator wav, On-Demand or Creator is the honest column. Developer is for a production app that needs 150 concurrent generations and 1,500 WebSocket connections, not for a producer who wants a nicer invoice line.
Two traps on that same pricing page, both dated 19 September 2026:
- The cost estimator says it is based on Realtime TTS-2 Flash rates, and that actual cost may be higher with Realtime TTS-2. If you quote the calculator without reading that sentence, you are quoting Flash as if it were TTS-2. Flash is the cheaper SKU. TTS-2 is the steered one.
- Overage is charged at the same rate as your plan. The $300 grant is not a hard ceiling that stops the meter. You can add credits or turn on auto-reload. Those extra dollars still spend at Developer rates. They are not a new product.
Batch synthesis is a third meter, not a fourth plan. Billing docs: the Batch API is 20% less per character than realtime, and tier discounts still apply, so the batch saving holds on every plan. On On-Demand TTS-2 that is $20 per million instead of $25, if you accept queue time. Batch does not turn $300 into a studio. It turns a pile of offline lines into a cheaper pile of characters.
Creator is the small paid rung people skip on the way to $300. $25/mo, $25 in credits, TTS-2 at $20/1M, Flash at $10/1M, 500 custom voices, 40,000 characters per Playground request, workspace creation and sharing, team management. For a content week that is a lot of voice file. It is still not an editor, and it is still not a seat in Versely.
Steering is the model, not the plan
The compare table prints Steering (Realtime TTS-2 only) as a feature of the TTS product, not as a Developer unlock. Inworld's TTS page, fetched the same day, draws the same line in the model grid: natural-language steering and non-verbal cues sit on Realtime TTS-2. Flash is the latency and cost SKU. You do not buy steering by paying $300. You buy steering by sending text to TTS-2 instead of Flash. You can do that on On-Demand at $25 per million characters.
That split is why the $300 myth is expensive. A team hears "steering" and "production" and jumps to Developer because the card says production applications. Steering was never the thing the card was selling. The card was selling rate, concurrency, and voice count. If you needed mid-line direction on a 2,000-character script, On-Demand already had the model. If you needed 150 generations in flight for a voice agent, Developer is the concurrency column. Those are different bills that share a brand name.
Flash exists on Inworld's public rate card. It is not a published Versely model URL, so it does not get a /models/ link here. On Inworld, Flash starts at $15/1M and falls to $8 on Developer and $7 on Growth. Use that SKU on Inworld's bill when first-audio latency or character volume is the constraint. Use TTS-2 when the constraint is delivery. Mixing those two reasons is how a quote sheet lists $8 and a steered take.
Inworld TTS 2 is the published Versely row as of 19 September 2026: text-to-audio, 2 credits per 1,000 characters, a voice file out. It is not a talking generate. There is no image input, no duration chip, no resolution. Style steering is why the row exists. The output is speech you still have to place. If a mouth has to match, that is a later lipsync pass, not a higher Inworld plan.
When the brief is emotion as the product rather than Inworld's character-steering habit, that is a different vendor bill. Hume Octave 2 emotional TTS is that page. Do not climb Inworld's ladder to buy Hume, and do not open Hume to dodge a $300 card you did not need.
A voice file is not a studio
Inworld is a realtime voice API with a playground. A studio, in production terms, is the place you can pick the model the job needs, caption it, fail over, and keep one credit pool when the voice row is the wrong row. Developer does not become that because the card says production. Developer becomes a cheaper TTS-2 character rate with a $300 prepaid wallet and agent-scale concurrency.
Versely's meter for this row is characters, not a vendor plan. Inworld TTS 2 is 2 credits per 1,000 characters on 19 September 2026. A 1,000-character script is 2 credits. A 10,000-character script is 20. That is a catalog number, not a dollar conversion, and it is not Inworld's $15. Do not turn Versely credits into USD. Recheck the model page on the day you quote; older posts in this corpus still say 5 credits, which is not the live card.
The door is AI text to speech. Run the row. Download the wav. Caption and lipsync after the take you will actually ship. If the week also needs a talking plate, a music bed, or mute autoplay text, those are other catalog rows on the same bill. They are not Inworld add-ons you unlock at $300.
Sign up at https://app.versely.studio/signup when the job is a voice file plus the rest of the slate, not a production-app rate card you will spend on one narrator.
The test is the same as every other vendor ladder. List the last ten audio files you shipped.
If eight of them were steered character reads and you were hitting a concurrent-generation wall, Developer is a rate conversation. You still caption somewhere else.
If eight of them were one-off voiceovers for short video, $300 is a prepaid bucket you will either underspend or drain on retries while you hunt a performance. On-Demand or Creator already sold you TTS-2. Flash already sold you the cheaper SKU. Steering was never the $300. The $300 was the discount on the characters, and only if you actually needed that many.
FAQ
Do I need the $300 Developer plan to use TTS-2?
No. On-Demand already lists Realtime TTS-2 at $25 per million characters, with up to 70 minutes included. Steering is a TTS-2 model feature, not a Developer unlock. Pay $300 when you want the $15 rate, the $300 credit grant, and the production concurrency column.
What does Creator's $25 actually buy?
On the 19 September 2026 card: $25 in credits, TTS-2 at $20 per million characters, Flash at $10, up to 33% off TTS and STT, 500 custom voices, a 40,000-character Playground paste, and team and workspace tools. It is a small prepaid rate cut. It is not a studio seat.
How does Versely bill Inworld TTS 2?
The published row is 2 credits per 1,000 characters. Output is a voice file, not a talking generate. Live numbers stay on Inworld TTS 2. This post does not convert those credits into dollars.