Every metered service, its unit, its markup, and the underlying provider rates.
Fifteen services are metered per unit. Each carries a markup over the provider's cost, and each requires an agreement before it runs on platform credentials.
The metered services
| Service | Unit | Billing | Markup |
|---|
| OpenAI | tokens | Per token | +20% |
| Anthropic Claude | tokens | Per token | +20% |
| Google Gemini | tokens | Per token | +20% |
| DeepSeek | tokens | Per token | +20% |
| Twilio SMS | message | Per message, varies by region | +50% |
| Twilio phone number — setup | number | $5.00 one-time | None |
| Twilio phone number — monthly | number | $5.00 per number / month | None |
| Twilio voice — inbound | minute | Provider cost + markup | +100% |
| Twilio voice — outbound | minute | Provider cost + markup | +100% |
| AI voice assistant (OpenAI realtime) | minute | ~$0.15 per minute | +100% |
| AI voice assistant (ElevenLabs) | minute | ~$0.15 per minute | +100% |
| Amazon SES email | email | Per email | +30% |
| Bulk email | email | Tiered, see below | None |
| Google Maps & address lookup | lookup | Tiered, see below | None |
| EasyPost shipping | shipment | Per shipment | +50% |
Markup applies only to platform-provided services
Bring your own credentials and the provider bills you directly — there is no platform cost to mark up. See How integrations work.
Tiered services
Two services are priced in tiers rather than a flat rate, and both pass through at cost.
Bulk email
| Volume | Per email |
|---|
| 0 – 10,000 | $0.0010 |
| 10,001 – 100,000 | $0.0007 |
| 100,001+ | $0.0004 |
Google Maps and address lookup
| Volume | Per lookup |
|---|
| First 1,000 | Free |
| 1,001+ | $0.005 |
AI model rates
These are the provider rates the platform bills against, before the 20% markup. Per million tokens.
OpenAI
| Model | Input | Output |
|---|
gpt-5.4 | $1.25 | $10.00 |
gpt-5.4-mini | $0.25 | $2.00 |
gpt-image-1.5 | $10.00 | $40.00 |
Anthropic
| Model | Input | Output |
|---|
claude-4.6-sonnet | $15.00 | $75.00 |
claude-4.7-opus | $15.00 | $75.00 |
Google Gemini
| Model | Input | Output |
|---|
gemini-3.1-pro-preview | $0.10 | $0.40 |
gemini-3-pro-image-preview | $2.50 | $10.00 |
DeepSeek
| Model | Input | Output |
|---|
deepseek-chat | $0.27 | $1.10 |
deepseek-reasoner | $0.55 | $2.19 |
Image generation
Image models bill per image rather than per token, in three quality tiers:
| Quality | Per image |
|---|
| Low | $0.01 |
| Medium | $0.04 |
| High | $0.17 |
Cached-input pricing is not modeled
Provider discounts that are not applied include cached-input pricing, off-peak rates and per-length tiers. A charge computed here can therefore be higher than a provider's best-case rate for the same tokens.
Working out a cost
provider cost = (input tokens × input rate) + (output tokens × output rate)
sales price = provider cost × (1 + markup)
deducted = sales price × your plan's spend rate
Worked example — 10,000 input and 2,000 output tokens on gpt-5.4:
input 10,000 × $1.25/1M = $0.0125
output 2,000 × $10.00/1M = $0.0200
provider cost = $0.0325
+20% markup = $0.0390
Reading rates programmatically
GET/service-pricing/catalogJWT
GET/service-pricing/{service}JWT
GET/service-pricing/activeJWT
