modelgrep

Anthropic: Claude Sonnet 4.5 vs Qwen: Qwen3.5-27B

Qwen: Qwen3.5-27B wins on more metrics (6 of 9), but the right pick depends on what you optimize for — see the breakdown below.

MetricAnthropic: Claude Sonnet 4.5Qwen: Qwen3.5-27B
Intelligence Index37.142.1
Coding Index33.534.9
GPQA Diamond73%86%
Design Arena Elo1242
Speed (tokens/sec)4560
Latency787ms919ms
Input price /M$3.00$0.195
Output price /M$15.00$1.56
Context window1M262K
CapabilitiesReasoningToolsJSONVisionReasoningToolsJSONVision