Top up a balance from your wallet and spend it on whatever model does the job. No plan, no card, no monthly floor you have to grow into.
Same request shape as the OpenAI API, so anything already pointed at that works here by changing two settings. The cost comes back attached to the response. You never have to go look it up.
That is the whole pricing model. Each model has a published rate and you are charged per request. Below is what the ones people actually use cost. Everything else is on the rate card.
| Model | Good for | Per request |
|---|---|---|
| gemini-3.7-flash | Bulk work where quality matters less than volume | 1.3 cr |
| claude-sonnet-5 | The default. Most people should start here | 7.4 cr |
| gpt-5.6-sol | Tool calling and anything agentic | 7.4 cr |
| claude-opus-5 | Long reasoning. Expensive, and usually overkill | 18 cr |
| flux-2-pro | Stills, priced per image not per token | 60 cr |
| kling-2.6 | Video. Budget for it | 290 cr |
Honest note. For most jobs Sonnet beats Opus on price and you will not notice the difference. We list Opus because sometimes you will.
Better you read this now than find out later.
Hold it and your platform fee drops, 25% off for every 1% of supply, to zero at 4%. That is the entire utility. It pays no yield, it is not a share of anything, and we would rather you ignored it than bought it expecting otherwise.
One billion supply on Solana, mint and freeze authority revoked at creation. Full tokenomics.