Trial
$5 included
Evaluate the complete API. No card required.
- All public models and tasks
- Roughly 7 hours of unified analysis
- Request-level usage history
Pricing
Pay for audio time. Nothing else.
An hour of unified analysis — transcript, emotion, and speaking style in one response — costs less than a coffee.
Billed by the measured second No platform fee $5 trial credit, no card
What does an hour of audio cost?
- Per audio hour
- $0.72
- Per audio minute
- $0.0120
Audio per month
hours
≈ $7.20/mo
≈ $7.20 per month
The $5 trial credit covers about 7 hours at this rate.
50% less than buying transcript, emotion, and style separately.
Published rates
USD per audio minute
| Model | Transcript | Emotion | Style | Affect | Unified analysis |
|---|---|---|---|---|---|
| Spectra 1 | $0.0045 | $0.0060 | $0.0060 | $0.0075 | $0.0090 |
| Spectra 2Preview | $0.0065 | $0.0080 | $0.0080 | $0.0100 | $0.0120 |
| Resonance | $0.0080 | $0.0080 | $0.0080 | $0.0100 | $0.0120 |
Pricing version 2026-07-11. Spectra 2 is a preview tier and is not yet serving inference traffic — use Spectra 1 or Resonance today. Every response reports its measured duration, applied rate, and estimated cost, so bills are never a surprise. Model identifiers, lifecycle status, and endpoints are listed in the model catalog; what each task returns is defined on the capabilities page.
Pay as you go
Published rates
Prepaid credits. No monthly fee, no minimum commitment.
- Credits never surprise-bill you
- Automatic recharge and balance alerts
- Receipts and full usage ledger
Scale
Custom agreement
Volume rates, invoicing, and support terms for larger teams.
- Volume-based rate review
- Contracted invoicing
- Priority implementation support
On-device
Models that ship inside your product.
oruk specializes in building speech models for edge deployments — phones, wearables, and small hardware. The same understanding as the API, compressed to run offline on the silicon you already have. In the cloud you pay for audio time; on your hardware you license per device, so audio never leaves it and cost never scales with minutes.
Evaluation
Free
Prove the model on your hardware before any commitment.
- Compiled evaluation build for your target device
- Accuracy and latency measured on your own audio
- Time-boxed license, no card required
Per-device license
Annual, per active device
One flat rate per device in the field. No per-minute metering.
- Rates step down with fleet size
- Model updates and re-tuning included
- Fully offline — audio never leaves the device
Custom build
Scoped engineering
A model built for your chipset, footprint, and task.
- Tuned to your memory, power, and latency budget
- Task- and domain-specific training
- Delivered as an SDK with integration support
On-device programs are scoped individually. Write to access@oruk.ai with your target hardware, task, and expected fleet size.