Seems like the AI Gateway internal credentials for DeepSeek (the provider itself) ran out of balance and it started using secondary providers (DeepInfra). The issue has been there for more than 2h. Those secondary providers are unacceptable in terms of latency for a product with production traffic… It’s a bit sad and frustrating, because I really like AI Gateway as a product… But if I cannot be certain that the best provider will be guaranteed because you guys ran out of of balance, it makes me loose trust in the system. I’d be very happy if you guys find a way that this doesn’t happen again. In the meantime, I had to add my own DeepSeek credentials in the BYOK section to be able to use the model. However, with this I just lost the benefit of the gateway, because I had to create a provider account and now need to maintain the balance of that other provider… Its really frustrating. I trust that your capable team will be able to solve this as soon as possible so it doesnt happen again.
