FAQ
For failed requests or a client that cannot connect, see Troubleshooting. For groups, account pools, inference credits, and other terms, see Core Concepts.
How do I get in touch about enterprise integration?
An enterprise group is now open. For enterprise integration, API integration, or partnership enquiries, join QQ group 794504445.
Why do model detection sites report a high fake rate?
Model detection sites and leaderboards are not fully reliable. Some suffer from paid rankings, sample bias, or opaque detection methods. Models called through aggregators such as OpenRouter are also frequently misclassified as "fake" by these tools.
Treat the results as a reference only, not as the sole basis for judging whether a model is genuine.
Which group covers image generation?
For image generation through the API, choose a group based on the model you need:
| Group | Example models |
|---|---|
ChatGPT Image | gpt-image-2, gpt-image-2.5-flare, gpt-image-2.5-sunburst |
Google Image | gemini-3.1-flash-image, nano-banana-pro |
Grok Image | grok-imagine-image-2.0 |
Select the corresponding group when creating an API key. Check the model marketplace for the full model list and current multipliers. In the web Creative Studio, select a model directly; no API key is needed.
A group without image generation returns 403 Image generation is not enabled for this group, see Error Codes.
How can I generate images?
- Web Creative Studio: Visit the TokenFlux Creative Studio directly. Generate images online with no API setup or client downloads required.
Image requests time out during a long wait
Use the current API address, then try Streaming Images. The client must support SSE and confirm success from the final image event.