New kimi-k2 on Fiction.liveBench
Posted by fictionlive@reddit | LocalLLaMA | View on Reddit | 2 comments
Posted by fictionlive@reddit | LocalLLaMA | View on Reddit | 2 comments
AppearanceHeavy6724@reddit
about typical for large Chinese models (except Qwen)
fictionlive@reddit (OP)
It's competitive or better than Opus-4 and GPT-5. It's way better than Sonnet-4. Also never mentioned this but it's clear GPT-5 is smaller than GPT-4o.