ChangeRadar
Benchmark

Moonshot Kimi — reported benchmarks

Moonshot AI · source page ↗ · last checked Oct 19, 2026, 12:05 AM

Reported benchmarks · Kimi K3

captured Oct 19, 2026, 12:05 AM
BenchmarkScore
GPQA Diamond93.5% · max reasoning effort
AIME 202596.1% · max reasoning effort
SWE-Bench Verified76.8% · max reasoning effort
Terminal-Bench 2.188.3% · max reasoning effort
FrontierSWE81.2% · max reasoning effort
DeepSWE67.5% · Kimi Code configuration
SWE-Marathon42.0% · max reasoning effort
Program Bench77.8% · max reasoning effort
BrowseComp91.2% · agentic configuration
GDPval-AA v21687 Elo · max reasoning effort
HLE-Full43.5% · no tools, max reasoning effort
ARC-AGI94.5% · ARC-AGI semi-private evaluation

Vendor-reported via automated web search — not independently verified. See the cited matrix on /models.

Change history