FAQ
Understand model capabilities, API requirements, and evidence status.
What can Model Atlas help me do?
Describe your task, filter AI models by capability, provider, and region, check official documentation, compare integrations, and export examples in JSON, cURL, Python, JavaScript, and other formats.
What is the difference between indexing, static validation, and live testing?
Indexing means model documentation was found. Static validation checks the minimal request's parameter structure. Live execution also depends on account permissions, API region, balance, and service availability. Static validation is not a successful API call or a quality guarantee.
How are original providers and third-party platforms distinguished?
The model developer and API platform are labeled separately. Several platforms may serve the same model. Use the selected channel's official endpoint, model ID, parameters, region, and pricing; do not mix them.
Does a Chinese provider always use a China-based API?
No. Company origin and API hosting region are independent. Chinese providers may offer international endpoints. China, Singapore, and other regions may have separate endpoints, keys, and model catalogs.
Do I need to provide an API key to choose a model?
Browsing and model selection do not require your API key. Exported examples contain placeholders that you fill in within your own environment. The website's task interpretation model is configured separately by the administrator.
Why can some models not export configurations?
Exports pause when an integration contract is incomplete, evidence is missing, documentation is outdated, the model is retired, or upstream changes await review. The model may remain listed without being presented as verified.
Will the website generate images, videos, or speech for me?
No. It provides discovery, selection, and integration guidance. Capability labels are not output guarantees; image generation alone does not establish precise preservation of original text or product pixels.