NewTrack your real AI spend with Usage Analytics

Know what every AI user actually costs you.

Price your AI product without the spreadsheet. Model token costs, margins and tiers across every major LLM provider, including the heavy users who quietly erase your margin.

Start modeling for free Try the calculator

Prices from 52 providers, synced every week.
No credit card required · Instant access

Pricing templates for every major model provider

  • OpenAI
  • Anthropic
  • Google
  • DeepSeek
  • Amazon
  • Meta
  • NVIDIA
  • Mistral
  • Perplexity
  • Qwen
  • xAI
  • Fireworks
Why Calcaas

Most AI products are priced on a guess. An average user, a spreadsheet, a margin that looks fine until the heavy users arrive. Calcaas counts every token, fee and customer before you commit to a price.

The catalog

2,808 models. Every one priced.

Each point is a model you can price today, coloured by the provider you buy it from. Text, image, video and audio, synced weekly, last on Mon, Sep 28.

  • OpenRouter476
  • Fireworks AI305
  • Fal AI210
  • Azure OpenAI208
  • AWS Bedrock150
  • DeepInfra134
  • Novita130
  • 45 more providers1,195
LLM pricing by providerCompare providersModel leaderboard
Features

See your true gross margin per AI user

From simple credit packs to per-token economics, Calcaas handles the math so you can focus on growth.

● Always current

Every model price, the week it changes

2,808 text, image, video and audio models, synced weekly from litellm, models.dev and the Vercel AI Gateway. You never type a price by hand.

  • Priced per host. The same model on fal and on the Vercel gateway can differ 2×. You see both.
  • Batch rates. Half-price batch tiers, where providers offer them.
  • Model finder. Set vision, tools, context and budget. Get the cheapest fit.
Price changes this weeksynced
Claude Sonnet 4.6Anthropic · per 1M tok
$3 / $15batch −50%
Veo 3 FastGoogle · per second
$0.15synced
Gemini 2.5 ProGoogle · per 1M tok
$1.25 / $10synced
Seedream 5Vercel AI Gateway · per image
$0.035new
last sync Mon, Sep 282,808 models
● Margin watch

Find the customer who costs you money

Averages hide your heaviest users. Calcaas models the whole spread, flags the tier where profit turns negative, and tells you what to change.

  • Usage segments. Light, typical and heavy users in one mix.
  • Exact break-even. The request count where each plan stops paying.
  • A fix, not a warning. Raise the price, cap usage, or switch models.
Growth plan · $19heavy-user check

4 users lose you $3.85 a month each

at 4,000 requests on GPT‑5, profit turns negative after 3,299
Option ARaise Growth to $24 → margin 76%
Option BCap at 3,000 requests, upsell Scale
Option CRoute to Gemini Flash → cost −82%
blended margin 71.4% → 74.9%option A

LLM pricing calculator

Switch between credit packs (images, video, audio) and per-token pricing in one click. Input and output rates handled automatically.

Cost simulation

Factor in fixed costs, payment fees, and what each free trial actually costs you.

Live currency rates

Model your pricing in 30+ currencies. Instantly update to the latest exchange rates to ensure your global margins hold up.

Saved scenarios

Stop rebuilding margin models from scratch. Save full pricing configs — cost components, packages, currency, billing cycle — and reload them on any device.

Billing cycles

Compare one-time, monthly and yearly models. Automatically calculate discounts and visualize the impact on MRR.

Quote history

Every PDF you export is recorded and re-downloadable, byte-for-byte stable, so a quote you sent in March still pulls the exact numbers months later.

Branded PDF quotes

One click and you have a client-ready PDF: package cards with profit and margin, full cost composition, your branding.

Model recommendation

Tell us your requirements — vision, reasoning, tool-calling, context window, budget — and get the cheapest model that fits.

Unit economics

Break down CAC, price per query and margin per user, so you know each customer pays for themselves.

● New · Usage Analytics

Now track what your AI tools actually cost

Model your pricing, then measure your real spend. Calcaas pulls your true Claude Code token usage — by model, device and day — into one private dashboard. Your prompts never leave your machine.

  • By model, tool and device, with daily cost trends.
  • Multi-device rollup and monthly budgets that warn you.
  • Free for everyone. Only numbers leave your machine, never prompts.
Explore Usage Analytics
calcaas.com/usage · last 30 dayslive
Tokens48.2M
Cost$214.60
Devices2
claude-opus-492%
claude-sonnet-454%
claude-haiku-421%
6 ways to see your spendfree
Every dollar

See where each dollar of a plan goes

Model cost, payment fees and profit, stacked to scale against the price. Then add a heavy user.

How it works

How to build profitable LLM pricing tiers

Go from guesswork to a data-backed pricing page in minutes. See the full walkthrough →

01

Define your costs

Pick models from the catalog, then add fixed operational costs and transaction fees.

Claude Sonnet 4.6$0.0096 / req
GPT Image 2$0.04 / img
Ops overhead$420 / mo
02

Count real tokens

Paste a real prompt. The token counter sizes input and output instead of guessing.

You are a pricing assistant. Summarize the call…

Input1,204 tok
03

Build your packages

Create unlimited tiers. Set a target margin and Calcaas finds the price, or set the price and see the margin.

$9100 req
$19600 req
$492k req
04

Analyze & stress test

Profitability, ROI and MRR in real time. Push costs up, add heavy users, run 1,000 random months.

P10 – P90 margin52% – 77%
05

Send the quote

Export a branded PDF in 30+ currencies. Every quote stays re-downloadable with its original numbers.

Q‑0142 · Growth€17.50 / mo
Gross margin66.3%
Statussent
Try it on your product

Price your AI product in 30 seconds

Fill in the sentence. Every number below is the real Calcaas math, no sign-up.

My product uses for requests per user a month, at about input and output tokens. I charge a month.

Drag a number sideways to scrub it, or type.

Margin78%
Profit per user$14.85
Break-even3,300requests
Where each $19 goes$0.0055 / request
Model $3.30Fees $0.85Profit $14.85
Above a 70% margin at this usageRoom to spare: 70% holds down to $14
Profit per user as usage growsdrag the dot
$19$10$0−$901,2502,5003,7505,000break-even 3,300$14.85

Profitable until 3,300 requests, 2,700 more than this user sends. Your heaviest users decide your real margin.

Same usage, other modelscost per user · margin at your price
Example list prices per 1M tokens. The app prices 1,500+ models across text, image, video and audio.Open the free LLM cost calculator →
Under the hood

Every input that moves your margin, in one model

The costs most founders forget are the ones that sink a plan. Calcaas asks for each one.

Model tokensin / out per requestUsage mixlight · typical · heavyPayment fees2.9% + 30¢, VATFixed costshosting, support, CACCurrency30+ live FX ratesCalcaaspricing engineList priceper plan, per currencyMargin & profitper user, fully loadedBreak-evenusers and requestsPDF quotebranded, re-downloadable
Before / after

Retire the pricing spreadsheet

Fragile formulas, guessed margins and manual FX conversions, swapped for one clear answer you can defend.

pricing_v7_FINAL(2).xlsx

“prices updated… March? check before launch”

calcaas · growth-plan
Starter · 100 req$9
Cost$2.31
Profit$6.13
Growth · 600 req$19
Cost$4.70
Profit$13.45
Scale · 2k req$49
Cost$15.40
Profit$31.88
prices synced weekly · 30+ currencies · fees includedmargin 69.4%
Free tools

Free AI cost tools

Estimate what any model costs, no login required, then turn that cost into customer pricing.

free

LLM Cost Calculator

Per-request, monthly and yearly cost for GPT, Claude, Gemini and hundreds of models.

Open tool free

Image Generation Cost

What DALL·E, Flux, Stable Diffusion and more cost per image at your volume.

Open tool free

Text-to-Speech Cost

Estimate ElevenLabs, OpenAI and Polly voice costs per clip and per month.

Open tool free

Video Generation Cost

Price Veo, Runway and more by clip length — per video, month and year.

Open tool free

AI Model Treemap

A visual map of leading LLMs by usage and capability, coloured by adoption trend.

Open tool

The full pricing app

Turn any of these costs into customer pricing, margins, packages and PDF quotes.

Get started free
Use cases

Built for every stage of growth

Founders & indie hackers

Make critical, data-driven pricing decisions in the early stages. Avoid the “too cheap” trap before you even launch.

Product managers

Price new AI features effectively. Manage profitability by simulating token usage against subscription revenue.

Finance & CFOs

Conduct detailed financial modeling. Forecast revenue, calculate break-even points, and report to stakeholders.

Browse pricing calculators by product type
Customers

Loved by SaaS builders

Join hundreds of founders optimizing their pricing strategy.

★★★★★
“I finally know what to charge for my GPT wrapper. Before Calcaas, I was just guessing and hoping I wouldn’t go broke on API fees.”
Latasha CesarIndie Developer · AI copywriting tool
★★★★★
“Discovered we were losing $8k/month on power users. The detailed unit economics breakdown saved our runway.”
Michael RossCTO · Enterprise knowledge base
★★★★★
“We raised prices 3x based on these models and churn actually dropped. Customers understood the value proposition better.”
Sarah JenkinsGrowth Lead · B2B SaaS analytics
★★★★★
“Better than our internal BI tool. The live multi-currency rates and real-time profit margin analysis are indispensable.”
David MillerCFO · FinTech platform
★★★★★
“We were underpricing our AI features by 40% before using Calcaas. The token cost simulator is a lifesaver.”
Alex ChenFounder · Generative art studio
★★★★★
“Finally, a tool that understands the complexity of hybrid pricing models. Saved weeks of manual spreadsheet headaches.”
Jessica WuProduct Manager · Video editing software
+35%

Average margin improvement, based on user-reported data.

1,200+

AI SaaS products modeled in Calcaas.

Pricing

Pay once. Price forever.

Transparent pricing for every stage: start free, then one payment for lifetime access.

Starter
$0forever

Validate your first AI product pricing strategy free.

  • 3 pricing packages
  • Basic margin analysis
  • USD currency only
  • Basic multi-modal support (image or text)
  • No unit economics
Start free
Pro most popular
$49lifetime · one payment

Everything serious founders need to avoid losing money on AI costs. Pays for itself with one saved pricing mistake.

  • Unlimited pricing packages
  • 30+ currencies with live exchange rates
  • Advanced unit economics (MRR, ROI, CAC, break-even)
  • 1,500+ AI model presets, synced weekly
  • Auto token counter (paste a prompt)
  • Branded PDF quote export with re-downloadable history
  • Saved scenarios, scenario diffing and stress testing
  • Margin, profit, fixed-price and unit-price strategies
Get lifetime access

Compare plans

Starter is for validating one pricing model. Pro is for iterating quickly — comparing 10+ price points and managing multiple billing strategies at once.

FeatureStarterPro
Pricing strategiesMargin onlyMargin, profit, fixed price, unit price
Service modelsImage or text (LLM)Image, video, text, hybrid
LLM providersCustom pricing only1,500+ models, weekly synced
Token countingManual entryAuto-count from prompt (tiktoken + community)
Video costing—Advanced (duration, audio, $/sec)
Package managementFixed (3 packages)Unlimited (add / remove)
CurrenciesUSD only30+, real-time rates, dual display
Billing cyclesOne-time onlyMonthly, yearly (with discounts), one-time
Financial metricsRevenue, cost, profitMRR, ROI, CAC, break-even
Cost simulationBasic op cost, allocation %Transaction fees, trial users
Custom included features—Add specific costs (e.g. support)
Smart pricing—Charm pricing (.95) & rounding
Comparison tools—Side-by-side package comparison
Saved scenarios—Unlimited save / load with rename
PDF quote export—Branded multi-package PDF, client-ready
Quote history—Re-download every past PDF export
SupportStandardPriority feature requests
Changelog

Product updates

We keep the models current with the latest AI pricing, and ship something new most weeks.

  1. v3.8.0new

    Usage Analytics — track your real AI spend

    Calcaas now closes the loop. Beyond modeling what your AI product should cost, you can track what your AI tools actually cost. Install the open-source calcaas-sync CLI, link each machine with a per-device key, and your dashboard fills with real Claude Code token usage — broken down by model, tool, and device, with daily trends, a live sync indicator, and monthly spend budgets. Only numeric aggregates ever leave your computer — never prompt content. Free for every signed-in account.

  2. v3.7.0

    Claude Fable 5 & Snowflake Cortex

    Fresh from the weekly catalog sync. Anthropic's Claude Fable 5 ($10/$50 per 1M) is now in the picker across Anthropic, Vertex, AWS Bedrock, and Azure AI. We've also added Snowflake Cortex as a new provider — 19 models including Cortex-hosted Claude, DeepSeek R1, Llama 3.1/3.3/4, Mistral Large2, and OpenAI GPT-4.1 and GPT-5, plus MiniMax M3, Kimi K2.6 on Azure, and Ministral 8B. The chat-model catalog now spans 1,653 models across 46 providers.

  3. v3.6.0

    Claude Opus 4.8 & June Pricing Refresh

    The latest weekly catalog sync is live. Claude Opus 4.8 is now in the picker across Anthropic, Vertex, and AWS Bedrock, alongside the GPT-5 family on Oracle Cloud, Gemini 3.1 Flash Lite, and new open-weight models like Step 3.7 Flash, Qwen3.7 Max, and Kimi K2.6. Several models also got cheaper — DeepSeek V4 Pro dropped 75% to $0.435/$0.87 per 1M, Qwen3.7 Max halved to $1.25/$3.75, plus lower cache-read rates across the Claude and GPT-4.1 families. The chat-model catalog now spans 1,626 models across 45+ providers.

  4. v3.5.0

    Model Recommendation Engine

    Stop picking models blind. Tell Calcaas what you need — vision, reasoning, tool-calling, a context-window floor, a monthly budget — and it surfaces the cheapest model that qualifies at your real usage, capabilities shown at a glance. The benchmark now tags every model with what it can do, filters by Vision, Reasoning, or Tools, and sorts by newest release. Every model also carries full capability data — input modalities, context window, knowledge cutoff, and release date — under cleaner, canonical names.

  5. v3.4.0

    GPT 5.4 on Azure AI

    Azure AI's GPT 5.4 family is now in the picker — GPT 5.4, GPT 5.4 Mini, and GPT 5.4 Nano, including dated snapshot variants — pulled in automatically by the weekly catalog sync. The chat-model catalog now spans 1,600+ models across 45+ providers.

See all 19 updates
From the blog

Guides, pricing teardowns & AI cost insights

Fresh thinking on token economics, model pricing and packaging your AI product.

View all articles

AI Pricing Is a Financial Minefield

0:0015:00
FAQ

Frequently asked questions

Everything you need to know about the platform and pricing models.

How do I price an AI product?

Pricing an AI product requires balancing fixed operational costs with variable token costs. Calcaas helps you model these variables to find a price point that ensures profitability while remaining competitive.

How much does GPT-4 cost per user?

It depends on usage. With Calcaas, you can input your expected input/output tokens per user to see exactly how much GPT-4 (or any other model) will cost you at scale.

Do you support Claude API pricing calculator logic?

Yes, we support all major LLM providers including Anthropic's Claude. You can easily switch between models to compare costs and margins.

What is Calcaas?

Calcaas is a high-fidelity SaaS pricing simulator designed for founders. It replaces fragile spreadsheets with specialized logic for credit-based, user-based, and complex LLM token economics.

How do I choose the right model?

Tell Calcaas your requirements — vision, reasoning, tool-calling, a minimum context window, and a monthly budget — and the recommendation engine surfaces the cheapest model that meets them at your real usage. The benchmark view tags every model with its capabilities, lets you filter by Vision, Reasoning, or Tools, and sorts by newest release, so you can compare options without guessing.

Can I use it for non-AI SaaS?

Absolutely. Calcaas is built for any usage-based SaaS. Whether you sell image generation, compute minutes, or just user seats, our flexible component builder handles it all.

Is the simulation truly accurate?

The calculations are mathematically precise based on your inputs. By factoring in often-ignored costs like transaction fees (3%+), operational overhead, and trial leakage, you get a much realistic view of your true gross margins.

What happens after I pay for Pro?

You get instant, lifetime access to all premium features and future updates. No recurring subscriptions—just a one-time tool for your arsenal.

Browse the full FAQ

1,500+ prices. One answer.

free to startPro is $49 once, never monthly

Stop pricing your AI product on a hunch.

Model your costs, find the heavy users and send a quote you can defend. Your first model takes about five minutes. No spreadsheet, no credit card.

Start modeling for free Try the calculator
  • No card
  • 1,500+ model prices
  • PDF quotes