// Plans

Build real products from day one.

Simple plans, real prices. Pro is $34/mo, Scale is $100/mo, MAX is $200/mo. Every plan starts with a 3-day free trial, card required.

// Pricing
Starter

Pro

$34/mo
5,250 credits / month

Build up to 7 products a month with full Midas AI on every screen.

  • Build up to 7 products a month
  • 5,250 build and chat credits a month
  • Builds run on Gemini, GPT-5.6 Luna or Claude Sonnet 5
  • Midas chat, 5 of the ten models: GLM-5.2, GPT-5.6 Luna, Gemini 3.5 Flash, Gemini 3.1 Pro and Claude Sonnet 5
  • Bank auto-connect (Plaid)
  • Full Midas AI on every screen
  • Super Products, Builder, Projects, Systems
  • Vault and Edge included
  • Starter workspace limits (see below)
Start for free
Power users

MAX

$200/mo
30,750 credits / month

Build up to 41 products a month, or bring your own key and build unlimited.

  • Build up to 41 products a month
  • 30,750 build and chat credits a month
  • Bring your own key, build unlimited
  • All build models: Gemini, GPT-5.6 Luna, Terra and Sol, Sonnet 5, Opus 5, Fable 5
  • Midas chat, all ten models: adds GPT-5.6 Sol and Claude Fable 5
  • Unlimited workspace limits
  • Official bookkeeping (Ledger)
  • Everything in Scale
Start for free

Every plan starts with a 3-day free trial, card required. Yearly billing saves up to 15%, charged once a year: Pro $348, Scale $1,020, MAX $2,040. Cancel anytime.

Live for every member today: the Product Builder and Midas chat. The rest is rolling out and unlocks for every member as it ships, at no extra cost.

// CreditsAI usage

Builds are priced by the model

Every build draws credits based on the model that runs it: Gemini is the 1x base, GPT-5.6 Luna is 2x, GPT-5.6 Terra is 3.5x, GPT-5.6 Sol is 5.5x, Claude Sonnet 5 is 6x, Claude Opus 5 is 10x, and Claude Fable 5 is 20x. In Midas chat, Sonnet is Pro and up, while Sol and Fable are MAX only. Chat is charged on what each message actually uses, so the figures below are typical, not fixed. Builds use the same models as chat: Pro gets Gemini, Luna and Sonnet 5, Scale adds Terra and Opus 5, and MAX unlocks the full ladder.

Midas messages have their own price

Chatting with Midas is separate from building. You are charged for what a message actually uses, so a short question costs a fraction of a long one. The tables below show what a typical message costs on each model.

Gemini goes furthest, save the big ones

Our own frameworks make every build great even on Gemini, the most affordable model. Reach for GPT-5.6 Luna, Terra or Sol, Sonnet, Opus or Fable on the jobs that truly need them, and your plan stretches a lot further. MAX users can also bring their own key and build unlimited.

// How credits work

Two simple ladders: builds and messages.

One pool of credits covers everything. Builds are priced by the model's build multiplier; Midas messages are charged on what each message actually uses. Not every job needs the most expensive model, so knowing which to pick stretches your plan much further. The full breakdown is right below.

// What your credits buy

Exactly what you can build and use.

One pool of credits covers everything. Builds are priced by the model's build multiplier (Gemini 1x, GPT-5.6 Luna 2x, GPT-5.6 Terra 3.5x, GPT-5.6 Sol 5.5x, Claude Sonnet 5 6x, Claude Opus 5 10x, Claude Fable 5 20x). Midas messages are charged on what each message actually uses, so their prices below are typical. Here is the full breakdown.

// Credit cost of each buildPer build

Simple web apps and web apps with a database are MAX only. Pro builds run on Gemini, GPT-5.6 Luna or Claude Sonnet 5. Scale adds GPT-5.6 Terra and Claude Opus 5. MAX adds GPT-5.6 Sol and Claude Fable 5.

// Credit cost of each Midas messagePer message

Attachments add a little: +1 per image, +2 per audio clip, +1 per PDF.

// Midas messages your pool buysPer month, if spent only on chat

Or any mix. Every build and message draws from the same monthly pool.

// Pro, $34/mo5,250 credits / month

With your monthly pool, the most you can build if you spend it all on one thing. Pro builds run on Gemini, GPT-5.6 Luna or Claude Sonnet 5.

Or spend it on Midas chat instead: about 1,312 GLM-5.2 messages, down to 525 Claude Sonnet 5 messages. Every build and message draws from the same 5,250 credits.

// Scale, $100/mo15,350 credits / month

Scale adds GPT-5.6 Terra and Claude Opus 5 builds alongside Gemini, GPT-5.6 Luna and Claude Sonnet 5. The most of each, if you spend the whole pool on it.

Pick the model per build. GPT-5.6 Luna and Terra cost more credits, so reach for them on the jobs that need them. On chat, Scale runs about 3,837 GLM-5.2 messages down to 959 Opus messages.

// MAX, $200/mo30,750 credits / month + your own key

MAX unlocks every build model. The most of each, if you spend the whole pool on it.

"none" means one build of that type costs more than the whole monthly pool. GPT-5.6 Sol and Claude Fable 5 builds are MAX only. On chat, MAX runs about 7,687 GLM-5.2 messages, down to 960 Fable 5 messages. Bring your own API key and build unlimited beyond the monthly pool, you are never cut off.

// Plan limits

How much you can run on each plan.

Credits power your AI. These limits are how many of each thing you can keep running at once. Pro is a focused starter workspace, Scale multiplies those limits about 8x to 10x, and MAX is unlimited.

// What you can runPer plan
// Advanced featuresBy plan
// Our pricing transparency

Why a cheaper model still ships a great product.

Most jobs do not need the most expensive model, because the hard part is already done for you. When Mythox builds, the frameworks do the heavy lifting, so your credits go further than they would anywhere else.

// Baked into every buildDone for you

Proven structure, security, and scale

Every build ships with security, scale, proven systems, state-of-the-art protection, app-store readiness, and future-proofing already built in, the same standards Mythox itself was built on. So a landing page or funnel comes out great on Gemini, the most affordable model.

Save the big models for the hard jobs

Complex logic, large databases, and deep reasoning are where GPT-5.6 Luna, Terra and Sol, Claude Sonnet 5, Opus 5 and Fable 5 earn their higher credit cost. Reach for them when a job truly needs it, not for everyday builds.

One honest pricing rule

Every action is priced by what the model costs us to run. Builds scale by a model multiplier (Gemini 1x, GPT-5.6 Luna 2x, Terra 3.5x, Sol 5.5x, Claude Sonnet 5 6x, Opus 5 10x, Fable 5 20x); Midas messages are charged on what each message actually uses, so a short question costs a fraction of a long one (a typical message runs from about 4 credits up to about 32). Use the cheaper models for most jobs and your plan stretches a lot further.

// Credit honestyFair billing

See the cost before you run anything

Every build and every message shows its credit cost up front, so nothing is ever charged by surprise.

A full credit history

Every credit you spend and every refund is logged in a running history you can read anytime.

Automatic refunds when a failure is ours

If a build fails on our side you are refunded automatically, and a repeat-failure guard stops you burning credits on the same broken job.

Optional monthly caps with alerts

Set a monthly spend cap if you want one and get alerts as you approach it. You stay in control of what you spend.

Top up any amount, credits never expire

Add credits in any whole-dollar amount whenever you need them, and they never expire.

// Honest by default

You always know what you are paying for.

Credits are only fair if you can see and control them. So every cost is shown before you run it, every charge and refund is logged, failures on our side are refunded automatically, and you can cap your spend, top up any amount, and keep credits that never expire.

// Live now

Start your free trial

Start with a 3-day free trial, card required. Cancel anytime. All prices in USD.