Unverified50% confidenceBenchmarkExact time

Grok 4.5在Terminal Bench基准测试中取得83.3%，基本与GPT 5.5打平，略低于Claude 5

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

7/11/2026

First Seen

Valid until: 10/9/2026

Sources

Grok 4.5编码实测：性能、速度与成本全面解析

bilibiliAI_Express7/10/2026

Related Claims

UnverifiedGrok 4.5能力尚未完全达到Claude 5或GPT 5.5的级别，但在能力、延迟和价格三者之间取得了平衡84% similar UnverifiedGrok 4.5在SWE-Bench Pro上取得64.7%，反超GPT Live 1和GPT 5.582% similar VerifiedGPT-5.4在SWE-Bench Pro测试中拿下57.7%准确率，超越了GPT-5.3-Codex的56.8%79% similar Unverified在编码能力上 Grok 4.5 与 Opus 4.8、GPT 5.5 基本持平，但响应速度最快77% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/488057

API

curl https://kongchang.com/api/v1/knowledge/claims/488057

MCP

get_claim(id=488057)