Usage & pricing
Organization usage is charged in credits. Amounts below come from the same pricing constants the product uses for billing. Text token prices for OpenAI models use the Standard tier by default unless your account is configured differently.
Credits and markup
| Item | Value |
|---|
| Credits per $1 USD | 10,000 |
| USD per credit | 0.0001 |
Storage
| Item | Value |
|---|
| Free tier | 100 GB (100,000,000,000 bytes) |
| Billed USD per GB-month (above free tier) | 0.10 |
| Credits per GB-month | 1,000 |
| Credits per GB-hour (accrual) | 1.3889 |
| Hours per billing month | 720 |
Analytics
| Item | Value |
|---|
| Free tier (per month) | 10 GB |
| Markup on GCP BigQuery $/GB (analytics only) | 7.5× |
| Billed USD per GB processed (above free tier) | 0.0375 |
| Credits per GB | 375 |
| Credits per GB-hour (accrual) | 0.5208 |
Product discovery (Google Custom Search)
Billed when product-discovery runs use live Google Custom Search to fetch AliExpress candidates. Each HTTP call to the API is one query; debits are posted immediately to the org credit balance (sub-kind product_discovery_google_cse). LLM steps in the same run use separate AI usage rows.
| Item | Value |
|---|
| Markup on Google Custom Search JSON API $/query (product discovery only) | 3× |
| Provider USD per query (reference) | 0.005 |
| Billed USD per query (client) | 0.015 |
| Credits per query (client total, rounded up) | 150 |
AI (OpenAI text tokens)
Unit: USD per 1 million tokens. “Billed” columns include the usage markup. Price table effective date: 2026-06-23.
Batch tier
| Model | Billed USD / 1M input | Billed USD / 1M cached input | Billed USD / 1M output |
|---|
| computer-use-preview | 7.50 | — | 30.00 |
| gpt-4.1 | 5.00 | — | 20.00 |
| gpt-4.1-mini | 1.00 | — | 4.00 |
| gpt-4.1-nano | 0.25 | — | 1.00 |
| gpt-4o | 6.25 | — | 25.00 |
| gpt-4o-2024-05-13 | 12.50 | — | 37.50 |
| gpt-4o-mini | 0.375 | — | 1.50 |
| gpt-5 | 3.125 | 0.3125 | 25.00 |
| gpt-5-mini | 0.625 | 0.0625 | 5.00 |
| gpt-5-nano | 0.125 | 0.0125 | 1.00 |
| gpt-5-pro | 37.50 | — | 300.00 |
| gpt-5.1 | 3.125 | 0.3125 | 25.00 |
| gpt-5.4-mini | 0.625 | 0.0625 | 5.00 |
| gpt-5.4-nano | 0.125 | 0.0125 | 1.00 |
| gpt-5.5 | 3.125 | 0.3125 | 25.00 |
| gpt-5.5-pro | 37.50 | — | 300.00 |
| o1 | 37.50 | — | 150.00 |
| o1-mini | 2.75 | — | 11.00 |
| o1-pro | 375.00 | — | 1,500.00 |
| o3 | 5.00 | — | 20.00 |
| o3-deep-research | 25.00 | — | 100.00 |
| o3-mini | 2.75 | — | 11.00 |
| o3-pro | 50.00 | — | 200.00 |
| o4-mini | 2.75 | — | 11.00 |
| o4-mini-deep-research | 5.00 | — | 20.00 |
Flex tier
| Model | Billed USD / 1M input | Billed USD / 1M cached input | Billed USD / 1M output |
|---|
| gpt-5 | 3.125 | 0.3125 | 25.00 |
| gpt-5-mini | 0.625 | 0.0625 | 5.00 |
| gpt-5-nano | 0.125 | 0.0125 | 1.00 |
| gpt-5.1 | 3.125 | 0.3125 | 25.00 |
| gpt-5.4-mini | 0.625 | 0.0625 | 5.00 |
| gpt-5.4-nano | 0.125 | 0.0125 | 1.00 |
| gpt-5.5 | 3.125 | 0.3125 | 25.00 |
| o3 | 5.00 | 1.25 | 20.00 |
| o4-mini | 2.75 | 0.69 | 11.00 |
Standard tier
| Model | Billed USD / 1M input | Billed USD / 1M cached input | Billed USD / 1M output |
|---|
| codex-mini-latest | 7.50 | 1.875 | 30.00 |
| computer-use-preview | 15.00 | — | 60.00 |
| gpt-4.1 | 10.00 | 2.50 | 40.00 |
| gpt-4.1-mini | 2.00 | 0.50 | 8.00 |
| gpt-4.1-nano | 0.50 | 0.125 | 2.00 |
| gpt-4o | 12.50 | 6.25 | 50.00 |
| gpt-4o-2024-05-13 | 25.00 | — | 75.00 |
| gpt-4o-audio-preview | 12.50 | — | 50.00 |
| gpt-4o-mini | 0.75 | 0.375 | 3.00 |
| gpt-4o-mini-audio-preview | 0.75 | — | 3.00 |
| gpt-4o-mini-realtime-preview | 3.00 | 1.50 | 12.00 |
| gpt-4o-mini-search-preview | 0.75 | — | 3.00 |
| gpt-4o-realtime-preview | 25.00 | 12.50 | 100.00 |
| gpt-4o-search-preview | 12.50 | — | 50.00 |
| gpt-5 | 6.25 | 0.625 | 50.00 |
| gpt-5-chat-latest | 6.25 | 0.625 | 50.00 |
| gpt-5-codex | 6.25 | 0.625 | 50.00 |
| gpt-5-mini | 1.25 | 0.125 | 10.00 |
| gpt-5-nano | 0.25 | 0.025 | 2.00 |
| gpt-5-pro | 75.00 | — | 600.00 |
| gpt-5-search-api | 6.25 | 0.625 | 50.00 |
| gpt-5.1 | 6.25 | 0.625 | 50.00 |
| gpt-5.1-chat-latest | 6.25 | 0.625 | 50.00 |
| gpt-5.1-codex | 6.25 | 0.625 | 50.00 |
| gpt-5.1-codex-max | 6.25 | 0.625 | 50.00 |
| gpt-5.1-codex-mini | 1.25 | 0.125 | 10.00 |
| gpt-5.4-mini | 1.25 | 0.125 | 10.00 |
| gpt-5.4-nano | 0.25 | 0.025 | 2.00 |
| gpt-5.5 | 6.25 | 0.625 | 50.00 |
| gpt-5.5-pro | 75.00 | — | 600.00 |
| gpt-audio | 12.50 | — | 50.00 |
| gpt-audio-mini | 3.00 | — | 12.00 |
| gpt-image-1 | 25.00 | 6.25 | 0.00 |
| gpt-image-1-mini | 10.00 | 1.00 | 0.00 |
| gpt-realtime | 20.00 | 2.00 | 80.00 |
| gpt-realtime-mini | 3.00 | 0.30 | 12.00 |
| o1 | 75.00 | 37.50 | 300.00 |
| o1-mini | 5.50 | 2.75 | 22.00 |
| o1-pro | 750.00 | — | 3,000.00 |
| o3 | 10.00 | 2.50 | 40.00 |
| o3-deep-research | 50.00 | 12.50 | 200.00 |
| o3-mini | 5.50 | 2.75 | 22.00 |
| o3-pro | 100.00 | — | 400.00 |
| o4-mini | 5.50 | 1.375 | 22.00 |
| o4-mini-deep-research | 10.00 | 2.50 | 40.00 |
Priority tier
| Model | Billed USD / 1M input | Billed USD / 1M cached input | Billed USD / 1M output |
|---|
| gpt-4.1 | 17.50 | 4.375 | 70.00 |
| gpt-4.1-mini | 3.50 | 0.875 | 14.00 |
| gpt-4.1-nano | 1.00 | 0.25 | 4.00 |
| gpt-4o | 21.25 | 10.625 | 85.00 |
| gpt-4o-2024-05-13 | 43.75 | — | 131.25 |
| gpt-4o-mini | 1.25 | 0.625 | 5.00 |
| gpt-5 | 12.50 | 1.25 | 100.00 |
| gpt-5-codex | 12.50 | 1.25 | 100.00 |
| gpt-5-mini | 2.25 | 0.225 | 18.00 |
| gpt-5.1 | 12.50 | 1.25 | 100.00 |
| gpt-5.1-codex | 12.50 | 1.25 | 100.00 |
| gpt-5.1-codex-max | 12.50 | 1.25 | 100.00 |
| gpt-5.4-mini | 2.25 | 0.225 | 18.00 |
| gpt-5.5 | 12.50 | 1.25 | 100.00 |
| o3 | 17.50 | 4.375 | 70.00 |
| o4-mini | 10.00 | 2.50 | 40.00 |
Other AI usage
When usage is billed from a single reported cost for a request, the billed amount is that cost multiplied by the same markup, then converted to credits (rounded up to the next whole credit). If the Vercel AI Gateway omits that cost but reports token counts, Cognis estimates provider USD from token-based fallback rates (see vercel-gateway-token-fallback-pricing.util in shared) and records gatewayCostSource: token_fallback on the usage row. Ecom AI content translation always attributes gateway translate calls to the requesting user or the org's billing user so usage is not skipped when a background job omits requestedByUserId.
