Cerebras launched Kimi K2.6 at a speed of one thousand tokens per second

Cerebras launched Kimi K2.6 at a speed of one thousand tokens per second. This trillion-parameter model is being launched at this speed for the first time; previously, the largest model from Cerebras was GLM 4.7 with 358B parameters. Unfortunately, this is currently only available to enterprise clients. By the way, the company also went public last week, raising $5.5 billion and is now valued at $60 billion. They are doing well, and I expect something bigger than Codex Spark from their collaboration with OpenAI.