modelgrep

MoonshotAI: Kimi K2.5 vs Qwen: Qwen3.5-27B

Qwen: Qwen3.5-27B wins on more metrics (5 of 8), but the right pick depends on what you optimize for — see the breakdown below.

MetricMoonshotAI: Kimi K2.5Qwen: Qwen3.5-27B
Intelligence Index37.337.2
Coding Index25.833.4
GPQA Diamond79%84%
Design Arena Elo1293
Speed (tokens/sec)8950
Latency211ms204ms
Input price /M$0.375$0.195
Output price /M$2.02$1.56
Context window262K262K
CapabilitiesReasoningToolsJSONVisionReasoningToolsJSONVision