modelgrep

Anthropic: Claude Sonnet 4 vs Qwen: Qwen3.5-9B

AAnthropic: Claude Sonnet 4anthropic/claude-sonnet-4VSQQwen: Qwen3.5-9Bqwen/qwen3.5-9b

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

MetricAnthropic: Claude Sonnet 4Qwen: Qwen3.5-9B
Intelligence Index33.0✓32.4
Coding Index30.6✓25.3
GPQA Diamond68%81%✓
Design Arena Elo1220✓—
Speed (tokens/sec)5068✓
Latency720ms362ms✓
Input price /M$3.00$0.100✓
Output price /M$15.00$0.150✓
Context window1M✓262K
CapabilitiesReasoningToolsVisionReasoningToolsJSONVision
Anthropic: Claude Sonnet 4 details →Qwen: Qwen3.5-9B details →