Pay What You Want

Drag one slider to set your monthly price. Where you land is your seat: you get its pool plus every cheaper pool beneath it, refilled each week. Nudge within a seat for more of that pool; cross a mark to step up to the pricier pool.

more Balcony tokens

$0$6$13$20$30$60$80

Drag the slider$26 / mo

Balcony's bar is $20, about 2.4M tokens a week. You're $6 over: about +2.1M more, so ~4.5M this week. Reach $30 to step up to Orchestra.

set it for real

First tier that bakes Sonnet 4.6 and Gemini 3.1 Pro into the weekly base. Mezzanine + Balcony pools come along for breadth.

◆Weekly Orchestra base — Claude Sonnet 4.6, Gemini 3.1 Pro
◆Plus weekly Mezzanine + Balcony (smaller than the Balcony tier)
◆PWYW slider lands bonus Orchestra tokens above the floor
◆150 image credits + 100 MB vectorization monthly — free image models cost no credits
◇No Producer/Exec base — cascade down from Orchestra to Balcony/Mezzanine

Models you can run

Weekly token pools

1.6M

2.6M

1.5M

Weekly tokens per pool. Higher pools can pay for lower-pool calls.

✦ Early adopter bonusup to 4.6× more

Orchestra: 1.5M vs 325K normally · 4.6×
Balcony: 2.6M vs 1.3M normally · 2.0×
Mezzanine: 1.6M vs 1.2M normally · 1.3×

Cascade reach

1.5M Orchestra tokens ≡ 1.5M Mezzanine · 1.5M Balcony
2.6M Balcony tokens ≡ 8.7M Mezzanine

Monthly extras

150 image credits · 100 MB memory vectorization

Each model draws its own weekly pool; a higher pool can cover lower-pool calls.

$30 more unlocks Producer

Slide within Orchestra Pit to add 313K Orchestra tokens per extra $1, up to $59.

Questions

What does "Pay What You Want" mean here?

Pick any price on the slider, from free to all-in. Where you land sets which models you can use and how much you can use them each week. Pay a little above a level's price and the extra simply adds more weekly usage at that level; reach the next level's price and you move up to it.

How does usage work?

Each plan comes with a weekly allowance at one or more quality levels: Mezzanine, Balcony, Orchestra, Producer, Exec, lowest to highest. Running a model draws from its level's allowance, sized to the length of each turn (longer messages use a little more). Everyday models draw from the big, inexpensive levels; flagship models draw from the smaller premium ones.

How are models assigned to a level?

By how capable the model is, nothing else. Fast, lightweight models sit at the lower levels; flagship reasoning models like Claude Opus sit at the top. Every model's level is shown before you run it, so there are no surprises.

What's cascade, and what happens if I run out at a level?

Allowances flow downward only: if a lower level runs dry, a higher-level balance can cover the call, never the other way around. And because cheaper models cost far less per token, that balance stretches as it cascades down, so one premium token can be worth several cheaper-model tokens. Your plan card's "cascade reach" line shows exactly what your balance is worth across the levels below it. Everything refills every Sunday, and you can top up any level at any time.

What is BYOK?

Bring Your Own Key. Every plan, including Free, lets you connect your own provider key (OpenRouter, NanoGPT, Anthropic, OpenAI, xAI, or a direct provider) and run entirely on your own account. Standing Room is the BYOK-first plan: the full platform, no platform-paid allowances.

Why don't the top plans include the cheap levels as a base?

Because cascade already covers them, and stretches doing it. Since cheaper models cost a fraction of premium ones per token, a premium balance spent on a cheap call goes a long way: a dollar at a high level buys far more cheap usage than the same dollar parked in a cheap base. So top plans put your money where you actually work and let cascade fill in everything below, for more usable capacity than a thin base spread across all five levels.

Is there a free tier?

Yes. Free gives you 3 chats and 1,000 messages total with BYOK enabled, so you can try the whole platform on your own key. Paid plans start at $6 (Standing Room: BYOK, no message caps); platform-paid allowances begin at Mezzanine ($13).

Curtain up.

Claim a seat. Write your scene.

Model and provider names (Anthropic and Claude, OpenAI, Google and Gemini, xAI, NanoGPT, OpenRouter, and others) are trademarks of their respective owners. RoleCall is an independent product and is not affiliated with, endorsed by, or sponsored by these providers.