Summary: TensorArt held two events and, when it came time to pay creators according to the specified rules, they added an extra requirement to evade making the payments. I will present the facts here. Evidence, communication, and rules are attached in a file.
Background: Recently, I participated in two events on the TensorArt platform: the SD3 event and the Hunyuan event. I published an SD3 Lora, an SD3 checkpoint for the SD3 event, one Hunyuan checkpoint, and 8 Hunyuan Loras. TensorArt promised to pay $20 for an SD3 Lora, $60 for an SD3 checkpoint, $150 for 8 Hunyuan Loras, and $100 for the Hunyuan checkpoint - a total of $330. On September 5, I received a notification on the TensorArt site regarding a transaction. However, 14 days later, I still have not received any payment to my bank.
Communication: I contacted TensorArt support, and they provided a PDF showing a single transaction of $12 - which I haven't received in my bank either. I’m unsure if this was an honest mistake or another layer of deception, but this is beside the point of this article. How does the promised $330 turn into $12? A TensorArt representative told me that only one of my models received 10 runs, which is why they were only paying for that one.
My response: The 10-run requirement was never mentioned in the event rules. This is a crucial condition that would have affected my decision on when to publish the models or whether to participate at all. I published my SD3 checkpoint on the last day of the event. In the attached SD3 conversation, the TensorArt representative initially claimed the model was published too late but then acknowledged it was on time, stating that payment would be made soon. Based on this confirmation, I decided to participate in the Hunyuan event as well. I published the models two days before the event ended, but they had deployment issues caused by TensorArt’s bugs. I had to open a support ticket to get the Hunyuan checkpoint deployment fixed, which took over a week. So it’s understandable that my models didn’t get 10 runs during the event. But, I'm not familiar enough with the platform to know how many runs models typically get or how to optimize for more runs. In any case, the 10-run requirement was never included in the event's original requirements.
SD3 and Hunyuan models are not popular at all. Creating Hunyuan models was especially difficult due to limited information and the need for an unconventional training toolchain.
I explained all this to the same TensorArt representative who previously promised that the SD3 payment was secured, but this time, they did not address any of these points in their response.
Legality: I love creating models, but I don’t appreciate being manipulated or misled by platforms. A platform can set whatever rules it wants for an event, but these must be clearly specified in the event details and not added after the event ends. Once a platform publishes an event, it has legal implications. Adding a 10-run requirement afterward is misleading and fraudulent. Without transparent rules, TensorArt could evade paying creators by retroactively adding specific unique requirements for each individual creator, which is clearly unfair and illegal.
What can be done: It would cost me a fortune to sue TensorArt if they were based in the US or Europe, but since they are likely located in China, it makes legal action nearly impossible. Therefore, it's crucial for model creators to understand that TensorArt cannot be trusted. It's also important for users to be aware of who they are supporting. The least I can do is expose the truth and let others decide for themselves whether they want to support a platform that misleads, lies to, and deceives content creators. I believe this issue may not be unique to my case, and I encourage other creators who have had similar experiences with TensorArt to come forward and share their stories.
My actions going forward: I will wait seven days for TensorArt’s response. If they respond, I will post an update. If they pay what was promised according to the event’s published terms, I will update as well. Otherwise, in seven days, I will delete my models and account from TensorArt and will recommend other creators and users do the same.
Attached are screenshots of my two communications with TensorArt that I saved (there was another communication about Hunyuan deployment issues that was closed and disappeared from Discord before I could capture it), event details, and screenshots of my models. Anyone can verify all the facts - rules, dates, communication, and promises - that I have presented as accurately as possible.
Edit 1:
Unfortunately, StableDiffusion subreddit decided to remove identical post to this one, because they think that informing the community about bad platform is "complaining".