modelgrep

Anthropic: Claude Sonnet 4 vs OpenAI: GPT-5.2

AAnthropic: Claude Sonnet 4anthropic/claude-sonnet-4VSOOpenAI: GPT-5.2openai/gpt-5.2

OpenAI: GPT-5.2 wins on more metrics (6 of 9), but the right pick depends on what you optimize for — see the breakdown below.

MetricAnthropic: Claude Sonnet 4OpenAI: GPT-5.2
Intelligence Index33.046.6✓
Coding Index30.644.2✓
GPQA Diamond68%86%✓
Design Arena Elo12201264✓
Speed (tokens/sec)50✓47
Latency720ms✓1.3s
Input price /M$3.00$1.75✓
Output price /M$15.00$14.00✓
Context window1M✓400K
CapabilitiesReasoningToolsVisionReasoningToolsJSONVision
Anthropic: Claude Sonnet 4 details →OpenAI: GPT-5.2 details →