Models

Use the model catalog to compare currently available models by capability, interface, pricing, and limits. API requests must use the exact Model ID shown in the catalog.

Choose a model

Task	Capabilities to check
Text chat and generation	Text input, Chat Completions, context, and output limits
Image understanding	Multimodal input, supported image formats, and size limits
Image generation	Image generation interface, dimensions, output method, and billing unit
Low-latency interaction	Streaming support and response latency

Image input capability does not mean that a model can generate images. Similar model names also do not guarantee identical interfaces, parameters, or output formats.

Check before calling a model

1. Copy the Model ID from the model card.
2. Confirm that the model supports the required interface.
3. Review its input types, output types, and parameter ranges.
4. Review context, maximum output, and media limits.
5. Confirm the pricing units for input, output, cache, images, or other usage.
6. Confirm that the current account and API key can access the model.

Compare pricing

Compare prices only after confirming that the billing units match. USD per one million tokens is a unit price, not the cost of one request. Image and other models may use different units. A public reference price is not the final charge for the current request.

Switch models

Keep the same test input and send a minimal request before moving traffic to a different model. Review:

• Response structure and business result.
• Supported parameters.
• Output quality and finish reason.
• Usage, latency, and cost.
• Application error handling.

Increase traffic only after the result is verified. Do not replace a production Model ID without testing it first.