GPT · Last 7 days
GPT-6.1 Sol
0–100 · higher means more positive community experience, not benchmark performance.
Updated 2026-10-02 14:32 UTC · 2026-09-25 – 2026-10-02 UTCRECENT EXPERIENCE
How the Experience Index is changing
30 days of rolling 7-day scores at each daily snapshot. The vertical scale adapts to the data. Missing or insufficient samples leave gaps; today is still updating.
Daily readings & sample sizes
| Date | Score | n | Scoring window (UTC) |
|---|---|---|---|
| 2026-10-02 | 56.8 | 9 | 2026-09-25T14:32:03+00:00 – 2026-10-02T14:32:03+00:00 |
| 2026-10-01 | 58.5 | 7 | 2026-09-24T23:57:02+00:00 – 2026-10-01T23:57:02+00:00 |
| 2026-09-30 | 66.6 | 5 | 2026-09-23T23:57:03+00:00 – 2026-09-30T23:57:03+00:00 |
Hacker News · 56.8 · 9 scored comments · Read platform reviews →
ONE MODEL, DIFFERENT COMMUNITIES
Across the communities
Scored separately for each platform, with different samples and audiences. Platform scores are not simply averaged; scored counts exclude quota-only comments.
Scores range from 0 to 100. Select a platform for its trend and reviews. Latest opinion time is not crawler health.
What people think
One user, after testing across 100 unsaturated coding and engineering environments, found 6.1-Sol clearly ahead of Opus 5.5 in these evaluations. One user on an x20 account running only GPT-6.1 Sol burned 10% of the weekly quota after just one day, calling the consumption rate horrific.
About the sample and scores
As of 2026-10-02 14:32:03 UTC, GPT-6.1 Sol in the GPT family has 376 explicitly attributed community comments in the last 7 days; its community score is 38.2/100. Available category scores include General text 49.2/100 (n=94); Coding 51.7/100 (n=10); Reasoning 70.0/100 (n=48). 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.
1 selected excerpts
Community reports about limits cannot establish official allowances.
Show original
因为这个模型带有循环，所以明文的思维链可能有一部分是假的。
Show original
很垃圾这个新模型做一个通宵 全错
Show original
让它改个函数实现，把接口改了，调用也全都改了
Show original
应该出个slow模式，20刀也可以猛蹬了
Show original
让它把几个软件工具的广告去掉并解锁付费功能，它是问都不问一句就哐哐地做啊
Comment context
We tested it across 100 unsaturated coding and engineering environments.
Show original
黑客能力被限制的死死的
Comment context
The model is a good model, I can only say, it's getting further and further from penetration testing. Now when I do penetration testing, I can only use the Luna model, hacker abilities are severely limited. Is it because we boarded the chatgpt ship, we can only research security defense from now on, and can never do penetration testing anymore? The average person's hacker dream might be over.
Machine translatedShow original
模型是好模型，我只能说，跟渗透测试，越来越远了。现在做渗透测试，我都只能使用Luna模型了，黑客能力被限制的死死的。是不是上了chatgpt这条贼船，以后只能研究安全防护了，再也不能搞渗透测试了。普通人的黑客梦，恐怕都要梦醒了。
No selected comments for this filter in the last 7 days.