technology
Kimi K3 Model Ranks Below Frontier AI Systems on LiveBench
The Kimi K3 model scored highest among open-source systems on the LiveBench evaluation but remained below several frontier models. The benchmark uses hidden questions designed to limit memorization.
Jul 17, 12:54 AM·1m read2 sources
BI
wccftech.comAudio version
Tap play to generate a narrated version.
Developing·Limited corroboration so far. This page will refresh as more sources emerge.
Kimi K3 achieved the top score among open-source models on LiveBench but placed behind Opus 4.8, Sol, and Fable. The benchmark contains hidden questions that make it difficult for models to rely on memorization. Kimi K3 performed below the three leading frontier models on those questions.
In practical use, Kimi K3 produces longer responses and processes queries more slowly than the frontier systems. Its cost per query matches that of Opus 4.8 for problems of comparable difficulty.
Transparency
How sources framed this
BI
BI
These outlets didn't split into competing frames — coverage was uniform.
CorroborationLimited · 2 sources
