modelgrep

Anthropic: Claude Opus 4.8 vs xAI: Grok 4.5

anthropic logoAnthropic: Claude Opus 4.8anthropic/claude-opus-4.8VSx-ai logoxAI: Grok 4.5x-ai/grok-4.5

xAI: Grok 4.5 wins on more metrics (5 of 9), but the right pick depends on what you optimize for — see the breakdown below.

MetricAnthropic: Claude Opus 4.8xAI: Grok 4.5
Intelligence Index55.7✓53.8
Coding Index74.3✓72.4
GPQA Diamond92%93%✓
Design Arena Elo1289✓—
Speed (tokens/sec)5774✓
Latency1.8s655ms✓
Input price /M$5.00$2.00✓
Output price /M$25.00$6.00✓
Context window1M✓500K
CapabilitiesReasoningToolsJSONVisionReasoningToolsJSONVision
Anthropic: Claude Opus 4.8 details →xAI: Grok 4.5 details →