I kept seeing people talk about open models but the company running the servers still controls everything
They can change the model or shut it down anytime But LightchainAI is its own blockchain built for AI Independent operators run models like - gpt-oss - Qwen3 - llama
Fixstars' domestic GPU version of DeepSeek costs 48 yen for input of 1 million tokens and 0.96 yen for reading cache.
It seems that input/output is not saved on the Fixstars server. I haven't touched it yet, but I would like to see the billing for one process that is difficult to send overseas.
You have reached the end of the archive
All of deepseek