OHGROK / Integration guide
Billing, balance & access
Keep model usage, group rates and account balance together.
On this page
Billing units
| Modality | Basis |
|---|---|
| Text | Token buckets or request counts, depending on the exact model and group. |
| Images | image-2 uses generated image counts; some Grok and GPT Image routes use request counts. |
| Video | Video billing tokens; resolution, duration and reference inputs affect estimates. |
| Image + text | Sum the separately priced text and image requests. |
Separate token usage
For token-priced models, keep uncached input, cache reads, five-minute cache writes, one-hour cache writes and output in non-overlapping buckets. If a provider reports cached tokens inside total input, subtract the included cache buckets before entering uncached input in the calculator. Use the returned usage and console record to determine which buckets are included.
When entering cache writes, select the usage format shown in the request record. OpenAI-format usage combines cache writes into one bucket; writes without a separate configured rate remain billable as ordinary input. Anthropic-format usage keeps five-minute and one-hour writes separate from ordinary input and applies the configured rate for each bucket. A missing rate and an explicit zero rate are different. The calculator’s initial selection is a preset, not proof of the format used by a route.
Only use the cache-write rates configured for that model and group. The displayed effective rates include the group multiplier once; do not multiply them again.
Context and group rules
Keep the full input length for each request when selecting a context tier. Group-specific expressions can have different thresholds or stay at one rate across lengths; do not apply a vendor comparison tier to the site price. A monthly token total does not determine individual request tiers.
Per-request pricing uses the number of billable requests. Generating several images in one request does not make its request count equal to the number of images. Check the model detail for the current unit and group price.
Balance and estimates
This website displays prices and budgets as dollar-marked site credits. One site credit, labeled $1 in the wallet, corresponds to ¥1 under the site’s 1:1 credit policy; the wallet label does not mean one US dollar. Model estimates use the exact group’s charge in these credits.
The cost comparison table shows all providers in US dollars, using the fixed comparison rate of US$1 = CNY 7. Official USD source quotes retain their original values. This comparison rate does not change the wallet’s 1:1 credit policy or the amount deducted for a request. Payment availability, fees and actual checkout amounts should be checked in the wallet; budget estimates include no unverified bonus.
Current access
Registration is open. The available top-up methods and promotions must be checked in the live wallet. Claude, GPT and Gemini are included in the current catalog; select an enabled model and a matching group. A listing does not guarantee current upstream capacity.
Reconcile usage
Keep model ID, group, time and request/task ID. Compare the response usage with the console records. Do not put secret keys into diagnostic logs.