PardnIO

1.2K posts

PardnIO banner
PardnIO

PardnIO

@pardnio

I built Agenvoy An AI agent that writes and repairs its own tools — and an MCP server let (Claude/Codex…) can build and share those tools too.

Taipei City, Taiwan Katılım Haziran 2026
409 Takip Edilen361 Takipçiler
歪理古德
歪理古德@goodming258·
你们是不是都用上了AI全自动评论了? 我看到很多评论都是牛头不对马嘴 一看就不是真人发的 只有我还在坚持码字手搓吗?
中文
14
0
4
437
PardnIO
PardnIO@pardnio·
@Y642873 好像有論文有分析這一個,有錢人與窮人的差別只在圈子
中文
0
0
0
3
谢梗哆
谢梗哆@Y642873·
穷人家的孩子 到底最缺的是什么 真的是钱吗?
中文
31
0
12
573
PardnIO retweetledi
Elon Musk
Elon Musk@elonmusk·
Once we have completed our review for security vulnerabilities, we will make the entire codebase of 𝕏 open source, with no exceptions. Moreover, we will invite third party reviewers to examine the system that is running to confirm that the open source code is what is running. Trust through total transparency is the only thing that should be believed.
English
4.2K
7K
63.4K
3.3M
PardnIO
PardnIO@pardnio·
@kkkm2238 無限重置真的很爽,但就重置卷要用時都很怕用完馬上又被重置,就浪費了
日本語
0
0
1
8
K
K@kkkm2238·
@pardnio 这两天我在做些游戏,可能耗的token很快,我以为这一周我搞不了了,但是一早给我重置了,我就觉得很舒服
中文
1
0
1
15
PardnIO
PardnIO@pardnio·
Unless we've created a highly well-known project before, few people will spend time or even money for a niche product. If we want to promote a product, we need to learn how to generate attention and build awareness first. What do you think?
English
1
0
3
70
PardnIO
PardnIO@pardnio·
@Smlay_2020 This is insane. This is exactly why it’s so hard to stop using GPT. If @grok can also offered a much higher weekly usage limit, it might actually have a chance to surpass GPT and Claude.
English
1
0
1
11
PardnIO
PardnIO@pardnio·
@AIShelterCN 其實你有看他的動作,動作都是在裝忙居多 浪費一堆不必要的 tiken 與時間在沒意義的動作
中文
1
0
1
15
AI避难所
AI避难所@AIShelterCN·
ai最大的缺点,就是不会假装自己很忙。。。
中文
8
0
6
695
PardnIO
PardnIO@pardnio·
Grok – Fastest and follows instructions well, but the weekly quota is too limited. Claude – Slower, but very reliable and low error rate. Codex – Slightly faster than Claude, but often does unnecessary things or ignores instructions, so the overall time savings are questionable. Senior engineer Need an AI that follows instructions and the workflow reliably, with faster responses. Grok > Claude > Codex Pure vibe coding Need an AI that can nail it from a single prompt. Don’t care about waiting—only care about getting it right on the first try. Claude > Grok / Codex (either is fine)
English
1
1
5
161
PardnIO
PardnIO@pardnio·
new feature `/usage` command with per-session usage will be release in Agenvoy v0.28.15. you can see how many tokens you've spent. is this would be helpful for you?
PardnIO tweet media
English
0
0
1
60
Divv Saxena
Divv Saxena@divvsaxena·
describe your product in 4 words. no links. just the 4 words. if the 4 words don't make someone curious, the landing page won't either.
English
16
0
14
784
Maxi Joe
Maxi Joe@ChainMaxi_Joe·
帐号关注数多出粉丝数500了,最近想清理下没回关的,找了一圈插件发现稳定性都很差,最后自己开发了个清理插件,在关注界面会给没回关的人一个红色标签,基本做到原生体验的程度
Maxi Joe tweet media
中文
14
0
10
2K
Ben Holmes
Ben Holmes@BHolmesDev·
Current model stack: - Grok 4.5 as default (grilling, planning, tasks around my computer) - GPT 5.6 Sol (high) for scoped implementation tasks - Fable *supervising* GPT 5.6 subagents for less defined feature work
English
17
3
187
13.5K
PardnIO
PardnIO@pardnio·
@ClassicOldSong 通常敢說,就應該是真的有做的 因為無法確保會不會有不長眼的員工爆露
中文
0
0
0
309
PardnIO
PardnIO@pardnio·
@cgnot996 光是速度快 + 足夠好的品質,就已經是生產級別的工具了 更別說 token 費用還低~(但比 Luna 高)
中文
0
0
0
58
铁柱AGI
铁柱AGI@cgnot996·
Grok 4.5 的强大被越来越多人发现了,他可能还不是最顶尖的模型,但是他的优点非常明显,能力大概在这个水平: Grok-4.5 > gpt-5.6-luna-max > glm-5.2 而且 Grok-4.5 速度比 gpt-5.6-luna-max 快太多了 如果对 Grok Build 有顾忌的话,可以用我开源的工具 GrokGo,把 Grok 4.5,图片生成,视频生成都接入 Codex 中使用!
诺鸭船长3@noahduck283

x.com/i/article/2076…

中文
9
3
35
8.3K