Opportunities › Cost estimation tools for LLM API usage › Proposed app
Tollgate
Estimate LLM API costs before you send, and cap spend per project
- Score
- 44
- confidence 46
- Evidence
- 3
- items on the page
- Rivals
- 1
- named in sources
- First version
- 30 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2023
- Emerging market · AI estimate
ScreensConcept
Desktop app: Developers paste prompts, review cost estimates across many parameters and set budgets in a detailed workflow at their desk before making API calls. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
New estimate
Estimated range $0.80 to $0.90$0.84
About this appAI proposal
Tollgate shows what an LLM API request will cost before you send it. Paste a prompt or a whole conversation, pick a model and settings, and see input and output cost as a range, not a guess after the fact.
It also shows how context grows across turns, so you can see the turn where a chat becomes expensive. Set a budget per project or key, get alerts as you approach it, and block calls that would go over.
Built for developers and small teams shipping LLM features who find out about spend from the monthly bill. Estimation is free; monitoring, alerts and limits are paid.
- Tagline
- Know what a prompt costs before you send it.
- Voice
- PlainPreciseCalm
- Why this name
- Tollgate: a gate you pass through and pay at. It stops a request, shows the price, and lets it through only if you agree.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Estimate before you send
Paste a prompt or message history, choose a model, and get an input and output cost range before any call is made.
No member yet
Context growth curve
Simulate a conversation over many turns and see cost per turn climb as history accumulates.
No member yet
Parameter what-ifs
Change model, max output tokens or cached context and compare the estimated price side by side.
No member yet
Budgets per project
Set daily or monthly caps per project or API key and see how much of each is used.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#F6F1E7Ledger Paper · surface#1E2A2BInk · text#0F5C5ADeep Teal · primary#E0902BAmber Cap · accent#B3392BSignal Red · data: over budget
Typography
Aa
Tollgate
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Know what a prompt costs before you send it.
PricingAI proposal
The material gives no competitor prices. It only says Calcis has a free tier with a usage limit and paid advanced features. These tiers are a starting point to test.
Free
Free
- Pre-send cost estimates
- Context growth simulation
- Shareable estimate links
Pro
$15 a month
- Spend monitoring
- Email and webhook alerts
- Budgets per project
- Pre-flight limit enforcement
Team
$49 a month
- Everything in Pro
- Shared budgets and members
- Per-key reporting
Would you pay Pro · $15 a month?
Your answer tells PWAS which prices people accept. It changes no score.
No answers yet.
What it could be worthEstimate
What a software business making this much sells for today, from real deals. Pick a monthly revenue.
If it makes
Worth about
$194,000
Sale range $132,000 to $263,000
$120,000 a year in revenue
From 87 software businesses listed and sold on Empire Flippers in the last year, revenue vetted by the broker: they sell for about 19× their monthly revenue, most between 13× and 26×.
Measured on profit, the middle deal is 32× monthly net profit: about 2.7 years of profit.
An estimate, not an offer or financial advice. A real price depends on profit, growth, churn, how much of the work is the owner's, and the buyer.
How it would growAI proposal
How the product would reach new customers through its own use, and how the first ones would be found. A plan to test, not a result.
- Built-in loop
- Shared outputs. Users share a link to an estimate or a conversation cost breakdown in pull requests, issues and team chats. Teammates who open it see the numbers and can run their own estimate.
- Do they become customers?
- Viewers are often developers who could become users, but a shared link may only be read once. Test how many link opens turn into an estimate run and then a signup.
- First customers
- Solo developers and teams of 2 to 10 building LLM features on Claude APIs, who have already had an unexpected bill.
- Where to reach them
- Hacker News: Post a Show HN with a free estimator and a write-up of how context buildup drives cost across turns.
- Developer communities: Answer cost questions in LLM developer forums and Discords with estimate links.
- Open source SDK: Publish a small free wrapper that prints estimates in the terminal and points to the paid limits.
- First milestone
- Ten teams with a budget limit active on a real API key, and three of them paying.
Why people already pay for this
- Public claimA developer incurred a $212 API charge from Claude Opus unexpectedly, partly due to context buildup across conversation turns.
- Public claimDevelopers search for tools to estimate LLM API costs before sending requests.
- Public claimCalcis offers a freemium model with free basic estimation tools and paid features.
Tech stack
Every proposal builds on the same PWA base, installable from the browser on any phone or computer. Dashed boxes are what this app adds.
- AppNext.js and TypeScript
- PWAManifest and service worker
- UITailwind CSS
- Data and authSupabase Postgres
- PaymentsStripe
- EmailResend
- HostingVercel
- This app · Token countingPer-model tokenizer libraries and a maintained price tableAccurate pre-send estimates need the right token count and current price for each model.
- This app · Request proxy / SDKLightweight proxy and thin client wrapperLimits must be checked before a call leaves, which needs a hook in the request path.
RisksAI inference
- TechnicalToken counts differ by model and provider, so estimates can be off. Output length is unknown before the call and must be shown as a range.
- PlatformProviders change prices and models often. The price table needs steady upkeep or estimates become wrong.
- CommercialA competitor already offers free estimation, and providers may add their own cost previews or spend caps.
- DataUsers must paste prompts or route traffic through a proxy. Many teams will not send sensitive prompts to a third party without local-only options.
- TechnicalA proxy in the request path adds latency and a failure point, so it needs a fail-open mode.
Everything marked AI proposal or AI inference was written by a model from the evidence on the opportunity page. The proposed name must be checked against trademark registers before it is used.