Why can full-capability models cost more?
Because flagship-grade access usually prioritizes better reasoning, richer capability, and more complete model behavior, its usage cost can be higher than lighter or reduced alternatives.
Transparent pricing
How full-capability model pricing works
Use this page to understand how full-capability model access is billed before you send large traffic or recharge for production workloads.
Your request type determines the cost
Text models are usually billed by input, output, and cached tokens. Image models are commonly billed by image count, size, and quality tier. Video, audio, and other media tasks are usually billed by duration, resolution, or completed task. Check the model catalog for the current unit price.
Three common billing methods
| Billing type | Typical requests | How cost is calculated | Watch closely |
|---|---|---|---|
| Token billing | Chat, reasoning, embeddings | Input and output are counted separately; some models price cache usage separately | Context, output length, cache hits |
| Per image / request | Image generation and editing | Image count combined with size, quality tier, or request type | Count, resolution, quality, edits |
| Media task | Video, audio, asynchronous generation | Duration, resolution, or successfully completed task | Duration, quality, status, retries |
Pricing guidance
- Review pricing guidance before sending production traffic to a new model.
- Different service plans or billing rules can apply different prices for the same model.
- Some models are billed per request or per generated asset instead of token usage.
Pricing FAQ
How do I know what a model will cost me?
Use this page as the pricing guide first, then check the live price shown in the product before heavy usage, because different models and request types can cost differently.
What should I check before sending large traffic or topping up?
Before large usage, confirm the current price, your balance or quota, and whether the target model supports the request type you plan to use.
For teams, not just self-serve
If pricing is tied to workflow risk, the next step is enterprise contact
If your team is comparing access stability, full-capability model behavior, usage records, and support expectations, a self-serve pricing table is only the first step. Use the enterprise page when you need to align on workload, budget, and rollout risk.
01
You run models inside Cursor, Cline, or an internal product.
02
You need clearer budget control, request logs, and a safer rollout path before scale.
03
You want to discuss maintained access resources, troubleshooting, or procurement workflow.