Unverified60% confidenceBenchmarkExact time
Grok 4.5在Terminal Bench 2.1上取得83.3%，超越Opus 4.8
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
7/11/2026
First Seen
Valid until: 10/9/2026
Sources
GPT Live全双工、Grok 4.5编程、Seedream 5.0图像：AI三赛道同步提速
bilibili养老院killer7/9/2026
Related Claims
UnverifiedGrok 4.5在Deep SWE-Bench 1.0上取得62.0%，明显领先Opus 4.882% similarUnverifiedGrok 4.5在SWE Bench Pro任务上的输出Token比Opus 4.8少4.2倍81% similarUnverifiedGrok 4.6的智能指数比Grok 4.5高5分，比Grok 4.3高23分75% similarUnverifiedGrok 4.6在Terminal Bench上的成绩从前代的15%跃升至26%74% similarUnverifiedGrok 4.6在Frontier Code基准从56.6提升至61.368% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/488370API
curl https://kongchang.com/api/v1/knowledge/claims/488370MCP
get_claim(id=488370)