Skip to main content
No. FreeModel uses prepaid USD balance. Add funds in Console > Billing and pay only for metered requests.
100 credits equal $1.00. For example, Seedance 2.0 Fast at 480p costs 7 credits per second, so a 5-second clip costs 35 credits.
Per one million tokens, with separate input, output, and cached-input rates on each model page. Reasoning tokens bill as output. GPT-5 prompts over 272,000 tokens bill the whole request at the long-context rate.
Per delivered image. gpt-image-2 costs $0.03 at low quality, $0.052 to $0.073 at medium, and $0.177 to $0.26 at high, depending on size. GPT Image 2.5 Ext costs $0.0085 at 1K, $0.014 at 2K, and $0.021 at 4K. gpt-image-2-all is not currently offered.
Seedance, Gemini Omni Flash 1.1, MiniMax H3, and Grok Imagine bill per output second, by resolution where the model offers several. Gemini Omni and Veo 3.1 bill a fixed price per request. Models that take reference or source video bill output plus reference seconds, which the gateway measures from the clips you send. The Playground shows the estimate before you submit.
No. A task that reaches the failed state releases its reservation and reports a zero final cost. Check the task and Console logs before resubmitting.
Referral cash and credit gifts are currently unavailable. New invite rewards are not issued. Credits already in your prepaid balance remain usable.
Use the live models page. Documentation pages record current rates, but the live catalog remains the source of truth.