Small businesses shouldn't have to compromise on AI quality due to unpredictable pricing.
That's why we are excited to work with Stripe's Billing for LLM tokens.
The Best Models Solve the Biggest Problems
Traditional software pricing doesn't work for AI. A flat $20/month subscription makes sense when every user costs roughly the same to serve. But with AI, one user might send 10 simple questions while another runs 500 deep research queries. The costs are wildly different.
This forces AI companies into an impossible choice: charge everyone a high flat fee (unfair to light users), cap usage aggressively (limiting what you can build), or lose money on power users.
We refused to accept any of those options.
What This Means for You
We worked with Stripe as their first design partner on Billing for LLM tokens—a new way to handle AI pricing that finally makes sense.
We did this to allow our customers to be able to pay for their Chipp usage based on the models and complexity they need.
Previously, you paid regardless of the business value and outcome that you were able to achieve. Now you can leverage the best tools to solve your customer's problems and get charged based on that specific consumption (and by proxy, the business value you're delivering).
What's more, every Chipp user gets 1/3rd of their subscription in monthly tokens - so a $99/mo plan has $33 worth of tokens every month as an example. That's a ton! Roughly 10m words a month or so on GPT-5 Mini or Gemini 2.5 Pro.
"We've been working with Chipp on a new product around how you pay for LLM tokens and how you bill your customers for LLM tokens, which has become a massive cost center for tons of new AI SaaS startups."
— Kimberly Hou, Engineering Manager, Stripe
Here's what this new model unlocks for you:
Use the right model for each job. Need a quick answer to "What's your return policy?" Use a fast, affordable model. Need to research competitors, analyze data, or provide personalized recommendations? Use the most powerful models available. You're not locked into one-size-fits-all. And you can use your monthly tokens on any model.
See exactly what you're paying for. Every query, every model, every cost—visible in your dashboard. No black boxes. No surprise bills. Set up alerts and payment thresholds so you have control.
Access frontier models from day one. GPT-5, Claude Sonnet and Opus, Gemini—the same models powering tools at the biggest companies are now available to a one-person shop building their first AI agent.
"Very quickly the cost starts stepping up. In that world where you have some users with $1 worth of usage and others with $100, a $20 fixed fee doesn't work so well anymore."
— Nick Dowse, Software Engineer, Stripe
Democratizing Access to the Best Tools for Small Businesses
This work makes AI accessible and practical for businesses that don't have engineering teams or six-figure AI budgets.
When a local boutique hotel can deploy the same quality AI concierge as a global chain—trained on their specific rooms, their neighborhood restaurants, their unique personality—that's the playing field we're trying to level.
Your AI agent should feel like your brand, not a generic chatbot. And you shouldn't need to choose between quality and affordability to make that happen.
Here’s how that looks in the real world.
How AI Model Mixes Work in the Real World
We're not talking theory. We're actively deploying this into hospitality as a cornerstone vertical, and the results speak for themselves.
The smart model mix in action:
A hotel guest asks "What time is checkout?" → Fast, efficient model. Instant response. Minimal cost.
A hotel guest asks "I'm celebrating my anniversary tonight—what's the best restaurant within walking distance with a romantic atmosphere, good wine list, and availability for two at 8pm?" → Powerful model with web search capabilities. Thoughtful, personalized recommendation. Worth the extra cost.
With usage-based pricing, hotels aren't forced to choose. They can handle high-volume simple queries affordably while delivering exceptional service on the moments that matter.
And we're pairing this with voice. As we give customers access to the best models through usage-based pricing, we're also rolling out voice agents that can answer calls, handle bookings, and provide concierge support—24/7, in any language.
See it in action: Check out how we're already working with hotels, and the massive ROI they're seeing only shortly into their engagement with Chipp: chipp.ai/hospitality
Why Transparent Pricing Improves AI Quality
We didn't use Stripe just for the aura (though yes, Stripe is an awesome company to work with). We worked together because this solves a real problem that was holding back our users.
When Kimberly and Nick from Stripe came to our launch event and walked through the integration live, it validated something we'd believed from the start: small businesses shouldn't have to compromise on AI quality because of unpredictable pricing.
When businesses can choose the power and price of each task, they can build sustainable business workflows. This is essential for the long-term impact of AI for actual business use.
"We started working with close design partners, people like Chipp AI, to build out a first version of usage-based billing—billing the customer based on how they're using your product, but specifically for LLM tokens."
— Nick Dowse, Software Engineer, Stripe
Together, Stripe and Chipp are building transparent pricing to power the next generation of AI use.
We gave feedback. They built. We tested. They refined. The result is a billing system that finally matches how AI actually works.
See the system in action:
What You Can Do Now
- Build with confidence. Pick a model to match your task. Use cheaper models for simple queries, and only use more expensive models for complex workflows.
- Check your usage. Head to Settings → Billing to see exactly which models you've used and what they cost.
- Learn more about pricing. Our Chipp Docs has an entire section on tokens. Visit this page to learn more and use our token estimator to see how many tokens you might use. Full pricing is shown on our Guide to Tokens post here.
- Build with us live. Thursday we are hosting a workshop on building your customer support agent. See the usage pricing in action and walk away with phone, internet, and WhatsApp support. Learn more and register here.
Chipp is one of Stripe's first Billing for LLM tokens alpha users. We're using this technology to bring transparent, usage-based AI pricing to small businesses.
