Real-world testing at a client site shows that Kimi's API call latency is still high in long-context scenarios. The $5B valuation reflects the market's trust in its compute investment. Have any of you deployed Kimi at scale recently? How are the thermal and power consumption metrics?
Physix Frontier
