Re:The Kimi K3 Moment
by staticman2 on Sunday July 19, 2026 @12:27AM
(#48964966)
For context you just asked me to read a document that starts with:
"Caveat: fully AI-generated research."
And that you quoted or paraphrased directly.
K3 reproduces Claude's internal model identifier when prompted, something which the real Claude models themselves do not emit. This is highly suggestive that K3 was trained on Claude metadata (API logs, tagged synthetic data), rather than Claude's chat outputs.
And it's well documented Chinese labs are paying good money for Claude metadata https://www.chinatalk.media/p/how-to-buy-cheap-claude-tokens...