Unverified50% confidenceOpinionExact time

推广文案宣称Grok 4.6性能吊打GPT-5.6与Claude，但这类绝对化表述缺乏公开可复现的基准测试支撑

1

Sources

50%

Confidence

Short-term (~14 days)

Relevance

9/12/2026

First Seen

Valid until: 9/26/2026

Sources

Cursor低价代充陷阱揭秘：共享账号池的风险与真相

Related Entities

Grokmodel GPTmodel Claudemodel

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79% similar UnverifiedGrok 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 578% similar Unverified推荐在Cursor中为Grok 4.6预设Claude Sonnet和GPT-4o作为备选模型，因其可用性相对稳定74% similar Unverified测试的两个项目此前已用Claude Opus 4.5加GPT-5.5的组合写过一轮74% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/910357

API

curl https://kongchang.com/api/v1/knowledge/claims/910357

MCP

get_claim(id=910357)