AI phone calls from
2 cents/minute.
Currently in private preview. Interested in a custom implementation? Reach out.
ThunderPhone v2 will launch publicly in Q4 2025.
Models
Model | Best for | Price from | Try it |
---|---|---|---|
⚡ Bolt | Complex tasks and high‑accuracy data entry. | $0.05/min | +1 (415) 639‑0815 |
✨ Spark | Simple use cases at the lowest possible cost. | $0.02/min | +1 (415) 639‑0860 |
Extra charges apply for premium voices (+1c/min), custom voices (+1c/min), and extra-long prompts.
What is ThunderPhone?
ThunderPhone is a phone AI platform optimized for the best possible performance‑to‑price ratio and ease of use.
To set up an automated phone number, just give us a prompt, pick a voice, and connect your integrations and VoIP.
ThunderPhone Bolt is the smartest model on the market, able to reliably handle even complex conversations with a single prompt. Bolt is also the most capable model on the market for accurate data entry over the phone.
ThunderPhone Spark is the cheapest model on the market - just 2 cents a minute! It's great for simple use cases.
Capabilities
- Dozens of languages supported out of the box.
- HIPAA & GDPR compliance pending October 2025.
- Easy bring‑your‑own software integrations and VoIP hookups.
- Outstanding turn‑taking performance, even in noisy environments.
Team & story
We’re a team of Y Combinator and Stanford AI Lab alumni that have been working on voice AI since 2024. Our CEO is Alex Kolchinski; our CTO is Alec Bell.
While building a call center AI system for a demanding enterprise client, we found that no existing phone AI platforms were good enough for their needs — especially accurate data entry. To solve their problems, we implemented a new system from scratch.
While doing so, we discovered ways to achieve high performance at low cost using an opinionated, integrated architecture blending commercial and open‑source models. After months of development, we’re now making our system publicly available as ThunderPhone.
We’re onboarding a few more customers this month for custom implementations — reach out if you’re interested. And stay tuned for our full public launch later this fall.