Why are models on Aitoearn so inexpensive?
Aitoearn is not only a software company. It also focuses on delivering operational outcomes for solo businesses.
The Aitoearn points you purchase, which cover token costs, are used to pay model providers such as Google, OpenAI, xAI, and Anthropic. We commit that our usage prices will not exceed the official prices of the corresponding models.
Because of our usage volume, we may receive discounts from model providers and pass those savings on to users. Our goal is to help more people earn with AI. Let’s use AI to Earn!
Limited-time models are not guaranteed to remain available.
Why is no one accepting my task?
Reasonable pricing, difficulty, and requirements make tasks easier for creators to complete.
What does the maximum payout mean for advertiser CPE and CPM tasks?
CPE and CPM tasks are billed by performance but can include a maximum payout. For example, you may set the reward at RMB 100 per 1,000 interactions with a maximum payout of RMB 50.
Creator A completes the task and receives 2,000 interactions. The performance-based amount is RMB 200, but because it exceeds the maximum payout, the creator receives RMB 50.
Creator B completes the task and receives 200 interactions. The performance-based amount is RMB 20, so the creator receives RMB 20 because it does not exceed the maximum payout.
For CPE calculation, a like or share counts as 1 interaction, a save counts as 3, and a comment counts as 5.
Why can a Douyin account with 10 followers accept a task that requires 100 followers?
Aitoearn currently calculates a creator’s total followers across all connected platforms. For example, a creator with 50 Douyin followers and 60 RedNote followers has 110 followers in total and can accept a task requiring 100 followers.
Follower restrictions are still a beta feature and are being improved.
Why does Seedance generation fail?
Seedance 2.0 is an AI video generation model from ByteDance with strict content requirements. Many generation failures are caused by the prompt. Revise the prompt and try again while considering the restrictions below.
1. Content Safety Restrictions
Content involving minors, sexual or pornographic material, identifiable real faces, copyrighted characters, political content, or extreme violence and gore is prohibited. PG-13 cinematic action, suspense-oriented horror, classical nudity clearly presented in an artistic context, and historical warfare presented with clear context may be allowed. The system applies prompt, model, and output moderation. Blocked content does not consume points.
2. Technical Limitations
Negative prompts are not supported. Instructions such as “do not” or “avoid” may have the opposite effect, so use positive descriptions. The recommended prompt length is 30-100 Chinese characters; content beyond 150 Chinese characters may be truncated. Each shot should contain only one primary movement instruction. Complex simultaneous movement across three axes and precise cinematography parameters are not supported. Text generation is unreliable and may produce garbled output, so add text during post-production.
3. Multimodal Reference Limits
Use the standard syntax Image 1 through Image 9, Video 1 through Video 3, and Audio 1 through Audio 3, and explain how each reference should be used. You can provide up to 9 images, 3 videos with a combined duration of 15 seconds or less, and 3 audio files with a combined duration of 15 seconds or less. The total number of files cannot exceed 12. References cannot contain real faces or copyrighted content.
4. Brand Content Restrictions
Logos, visual identities, signature products from well-known brands, and competitor brand content are blocked. Your own brand can be introduced only through an uploaded reference image, not through text alone. A character in the video may say the brand name, but voice-over use is not recommended. The safest workflow is to generate a brand-free video and add brand elements during post-production.
5. Real-Person Restrictions
Do not upload reference images or videos containing real faces, generate identifiable real people through text prompts, or generate images of minors. Supported approaches include using the platform’s virtual avatar library, creating a personal digital avatar through liveness verification for personal non-commercial use, reworking a face-containing video generated under the same account, or applying for an enterprise allowlist.
6. Best Practices
Use the prompt structure subject → action → environment → camera → style → constraints. Describe the visible scene like a director instead of using subjective impressions. Change only one variable per iteration so problems are easier to identify. Use reference images when character consistency is important.Last modified on July 15, 2026
