← All models

Claude · Last 7 days

Claude Sonnet 5.5

0–100 · higher means more positive community experience, not benchmark performance.

Updated 2026-10-01 09:57 UTC · 2026-09-24 – 2026-10-01 UTC
Experience score65.4/100113 comments in 7 days

What people think

One user recommends using Claude Sonnet 5.5 high as the implementer in a coding workflow supervised by Opus 5.5. One user states they wouldn't trust Sonnet to provide any architecture or design advice.

About the sample and scores

As of 2026-10-01 10:42:02 UTC, Claude Sonnet 5.5 in the Claude family has 113 explicitly attributed community comments in the last 7 days; its community score is 65.4/100. Available category scores include General text 67.2/100 (n=55); Coding 75.0/100 (n=8); Reasoning 63.6/100 (n=23). Scoring method and sources

Community reviews

Selected comments from the last 7 days. The balance of excerpts does not represent the share of positive reviews.

10 selected excerpts

Community reports about limits cannot establish official allowances.

NegativeReasoning

“I wouldn't trust sonnet to do give me any architecture/design advices.”

NegativeGeneral text

“This thing is inferior to Opus 5.5 in every way, and it's also learned the essence of how domestic small-parameter exam-prep model makers inflate their scores with flashy 'thundering deep thinking'.”

Machine translated
Show original

这玩意儿哪里都不如Opus 5.5,也是给A畜学到国产小参数做题家模型雷霆大思考刷分的精髓了

NegativeGeneral text

“If you're using the API, then it's worthless. Set the thinking level too low and you're an idiot, set it too high and the costs approach Opus-level.”

Machine translated
Show original

如果你是用API,那就是路边一条,思考等级定低了就是傻狗,定高了成本就朝Opus那边去了

NegativeReasoning

“Relying on unlimited extended thinking to achieve so-called higher intelligence feels like it might not be the right direction.”

Machine translated
Show original

依靠无限制拉长思考来换取所谓的更高智商,感觉可能不是正确方向

NegativeReasoning

“Follow up, found hallucinations”

Machine translated
Show original

追问,发现有幻觉

Comment context
From the same comment ↗

I discussed a JEV-related question and found that Company A's models now have a certain characteristic.

Machine translated
Show original

我拿一个跟JEV相关的问题讨论了一下,发现A家的模型现在有个特点。

NegativeUsage limits

“Company A's account is "delicate" — every now and then it completely blocks your mainland China IP.”

Machine translated
Show original

A社的账号"娇贵"啊,隔三岔五给你大陆ip全干挺了

NegativeSpeed & latency

“Thinking uses more tokens, making it slower”

Machine translated
Show original

思考用更多token导致更慢

NegativeSpeed & latency

“In the same thinking tier, Sonnet 5.5 consumes more tokens and is less efficient than Opus”

Machine translated
Show original

同思考档位的sonnet 5.5 比 opus 消耗token更多,效率更低

How's your AI experience today?