Grok

66.2 Sharper than usual experience score ▲ 7.1 Confidence: Medium

128 opinions tracked this week · 7-day rolling window

30-day experience score

Version breakdown

Only comments that explicitly name a version contribute to its score.

Grok 4.5	70.3+4.1	n=25	Confidence: Medium
Grok 4 Fast	mentions only	n=1	Confidence: Insufficient
Grok Build 0.1	mentions only	n=1	Confidence: Insufficient

101 generic family mentions are included in the family score, not any version score.

By experience dimension

General text	48	n=45
Coding	72	n=15
Reasoning	78	n=19
Image / vision	52	n=5
Roleplay / creative	51	n=8
Safety & refusals	40	n=10
Speed & latency	78	n=26
Local deploy	not enough data	n=0

Top positive drivers

speed quality Speed & latency ×20
strong reasoning Reasoning ×15
competitor flight General text ×6
creative surprise General text ×3
problem solved General text ×3

Top negative drivers

competitor flight General text ×9
safety overreach Safety & refusals ×4
hallucination Reasoning ×2
roleplay limit Roleplay / creative ×2
persona behavior General text ×2

What users are saying

🟢 “推理性能相比前代直接翻倍，空降到第一梯队水平，而且无论速度，成本都具有较大优势”
via zhihu · reasoning · zh

⚪ “At least the tok/s is fast so far, I like that. It seems like it lacks... depth?”
via reddit · speed_latency · en

🔴 “I refuse to allow a nazi to win the AI race.”
via reddit · safety_refusal · en

🔴 “体验下来就是速度确实飞快，就看着他疯狂刷屏，很快阿，就三分钟，git 里就堆了 200 多修改。但是说实话写的也不怎么样，发给肥波直接打回，硬编码走捷径问题太严重。甲的问题了，当年的 Grok 是以反 Woke、嘴臭、无过滤宣传的，然后甲就越来越厚。直接到这一代，直接从精神小妹变成了毫无灵性的苦逼打工人。破甲难度大幅提升就算了，写出来的东西很干、很机械，还容易输出 markdown 列表。对话中还会吐出 [System Note: ...] 这种设定词，非常的 OOC，非常的”
via zhihu · roleplay_creative · zh