This is just terrible. They proxy everything to ***. The generation speed is three times lower than on my own server with an RTX 5080, where both LLM and RAG embeddings, and even MCP work simultaneously. A response to a prompt with 6819 tokens takes a whole 35 seconds. And I have to pay for this?! It's really a disaster! And after contacting support on Telegram, they deleted the chat and blocked me. What a miracle.
