Pricing

No free tier. No freemium. No "contact sales for enterprise."
Two tiers. Transparent overage. The benchmark determines whether it is worth it.

Tier 1 — Developer
$500 / month

For developers building production voice agents and evaluating behavioral intent classification on real traffic.

  • • Full API access
  • • 50,000 classified audio segments per month
  • • Standard SLA
  • • Access to the benchmark dataset and methodology
  • • Integration guides for Vapi, Retell, LiveKit, Pipecat
Overage: $10 per 1,000 segments above limit.
Tier 2 — Production
$2,000 / month

For teams running voice agents at production scale where false interruptions are a documented cost line.

  • • Full API access
  • • 250,000 classified audio segments per month
  • • 99.9% uptime SLA with credits
  • • Dedicated support channel
  • • Priority access to new classification categories
  • • Integration support for custom stacks
Overage: $8 per 1,000 segments above limit.
THE MATH

VAD fires on approximately two-thirds of backchannels. Every false interrupt triggers an LLM API call that is then discarded.

On a 1,000-call production sample:

  • Estimated wasted API calls per conversation: 4.2
  • LLM cost per call at current market rates: $0.015
  • Monthly savings from eliminating false interrupts: measurable at scale

At most token pricing levels, the production tier pays for itself rapidly. The benchmark methodology is public. Run the numbers on your own traffic before you sign up.