kucinglapar

Cuma pengamat sosial media

[Ahmad Awais @MrAhmadAwais]

GLM 5.2 is not a better model than any for frontend.

Combine it with /design in Command Code - and you get the best design workflow.

> GLM-5.2 (Max) ranks higher than Claude Opus 4.8 (Thinking) on Code Arena, where Frontend coding tasks are being voted on by the community.

[yunus @0xyunss]

selama ini banyak yang nanya kenapa model AI China bisa jauh lebih murah dan jawaban "subsidi pemerintah" itu terlalu simpel.

ini lebih ke soal dua strategi yang beda fundamental.

model Barat main di "jadi yang paling pintar" bangun model sebesar mungkin, kejar benchmark tertinggi, ambil segmen enterprise yang rela bayar mahal. wajar kalau biaya per token-nya tinggi, karena overhead-nya juga besar.

model China main di "cukup pintar, tapi murah" optimasi engineering lebih ekstrem, target developer dan startup di Asia yang sangat sensitif harga, rela margin tipis buat rebut market share dulu.

dan grey market yang disebut di tweet ini? itu salah satu caranya dapat reasoning traces berkualitas tinggi dari model frontier sekaligus dapat revenue dari resell API yang "dipinjam".

yang menarik buat kita sebagai builder di Indonesia: buat kebanyakan use case sehari-hari, "good enough" itu sudah cukup. gua sendiri mix keduanya tergantung task-nya.

tapi kalau kalian pernah beli API murah dari third party yang harganya gak masuk akal sekarang udah tau dari mana asalnya.

[Command Code @CommandCodeAI]

Command Code is exceptional at understanding codebases.

We gave it an entire Next.js + TypeScript storefront and it understood the entire codebase in 3 prompts.

- Reads the schema and maps the structure
- Surfaces the patterns holding it together
- Traces a real flow, request to paid, file by file

Drop any unfamiliar codebase and get onboarded quickly.

[Hudya @perogeremmer]

Saya kasih tipsnya biar kalian hemat.

Opencode Go - $10

Dorong pake model Deepseek v4 flash atau Mimo v2.5 flash lebih banyak, trust me, in simple case dia udah cukup helpful. Apalagi buat case mahasiswa dan fresh grad.

Kalo bener-bener yang harus mikir, naikin variant ke high atau max, atau switch ke Deepseek v4 atau Mimo v2.5 pro.

Fallback, topup 5$ di Deepseek API langsung, kalo misal dah habis yang opencode go, pake fallback dan fokus ke deepseek v4. Ini saya pake buat AionUI, so far 5$ belom habis karena ya saya pake flash mulu.

Command Code - 1$

Lagi-lagi dorong pake Mimo dan Deepsek v4 flash. Udah worth it banget walau saya gak suka sih CLI macam gini, lebih suka OpenCode.

Xiaomi Mimo Token Plan - 6$

Lumayan bisa multimodal, dan Mimo v2.5 tuh udah bagus buat coding.

---

Kuncinya di bagaimana lu bisa jelasin dengan baik ke AI, dan gak ngarep AI tahu isi kepala lu.

So far saya gapernah menghabiskan 1 juta rupiah untuk ngoding, dan tetap bisa produktif kok 😁

[painn @_0xpainn]

Found a way to run GLM-5.2 for FREE with no daily limits through Cloudflare Workers AI 😳

No credit card. No monthly bills. Just one API token and you get solid usage.

What GLM-5.2 on Cloudflare unlocks:
-Strong agentic coding performance
-262k context window
-Tool use and function calling
-OpenAI-compatible endpoint
-Works with OpenCode, Cursor, Aider, Hermes, and more

What this replaces:
-Paid frontier model usage for daily coding agents
-Expensive API costs just to test workflows

How to grab it (5 min):

1. Create a free Cloudflare account

2. Go to Workers AI and copy your Account ID

3. Create an API Token

4. Add the provider in OpenCode with this base URL:
https://t.co/I06wzl3TnF

5. Select model: @cf/zai-org/glm-5.2

6. Start running agents

Works in OpenCode, Cursor, Aider, Claude Code, Hermes Agent, and any OpenAI-compatible tool.

Your buddy pays $20–200/mo for coding agents. You get strong performance for $0.

Bookmark this before Cloudflare changes the free tier.

[Tony Simons @tonysimons_]

$10/month for (essentially) unlimited DeepSeek V4 Flash.

That's the @opencode Go plan.

160K stars on GitHub, 7.5M monthly devs.

DSV4F is genuinely capable and stupidly cost-efficient.

The Go plan gives you a ton of usage on it -- plus generous access to GLM 5.2, Kimi K2.7, MiniMax M3, and more.

First month is $5 to try it.

No lock-in.
Cancel any time.
Use it with OpenCode, Hermes Agent, or any other agent.

Not affiliated with these guys. Just a happy user.

If you sign up with my link, we both get $5 in extra usage. That goes a long way on DSV4F. 🤘🏻

https://t.co/dUebgpjLDt

[Ahmad Awais @MrAhmadAwais]

Doubled the rate limits for everyone!!

[kucinglapar @kucinglaparssss]

@MrAhmadAwais Good feature addition. Nice it now get 5 hour limit. What model dan suited for 5 hour window? GLM / Minimax / Kimi will drain fast that limit. How about introduce "Auto Mode", that auto select model based on prompt request and how complicated that request?

[Command Code @CommandCodeAI]

We are doubling the usage limits to:

5-hour limit — 30% of your monthly credits
Weekly limit — 60% of your monthly credits

For all plans Go/Pro/Max.
Provider and extra usage has no limit.

We've also added major capacity, new providers, and a lot more GPUs to our self hosted infra.

- Most models now run 5x to 8x faster
- p50 TTFT: 5s → 965ms

If Command Code felt slow the last few weeks, give it another go. It's fast now.

Bigger plan dropping soon too. If you like the Go plan, this one's gonna hit home for y'all.

[titik kumpul wong mumet 🧭 @esjusalpuket]

kesibukan wni hari ini

[程叙AI. @chengzhx]

🔥 OpenCode最近太火了吗？之前执行飞快的DeepSeek V4 Flash 免费模型，今天使用起来怎么开始排队了啊😅😅 @opencode

[Fabio Vedovelli @vedovelli74]

Primeira vez usando GLM 5.2 via OpenCode.
Refactoring de médio porte numa codebase GoLang.

- É bem mais rápido do que o Opus 4.8 🤩

- É muito eficiente no uso de tokens 💵

- Acertou de primeira a avaliação do que era necessário refatorar (posteriormente validado por Codex e Opus) ✅

- Qualidade de entrega superior ao Opus 4.8 🏆

Estou satisfeito. Muito!

[Lukasz Olejnik @lukOlejnik]

GLM 5.2 looks like the first Chinese open-weight model to have entered the frontier tier in agentic cybersecurity tasks. Or to have copied it very effectively. The mistakes are interesting. GLM 5.2 gets things wrong in similar ways to Opus and GPT-5.5. That could mean a breakthrough. It could also mean someone trained it on answers from frontier models (distillation). In practice, the difference may be irrelevant to the user. An open-weight model reaches the level of the best closed models, at roughly half the cost of Opus.

https://t.co/hoLQiWa9Ll

[Devin Desktop @devindesktop]

Kimi K2.7 Code and GLM 5.2 are available in Devin Desktop and CLI

Both perform strongly on FrontierCode Extended, our benchmark for real-world engineering tasks:
GLM 5.2: 43.0%
Kimi K2.7 Code: 39.5%

Pro/Max/Teams users can try both models for free until July 5

[kucinglapar @kucinglaparssss]

@CommandCodeAI I'm trying your commandcode CLI with Github Spec Kit. And I found Deepseek give better result and reasoning.

[kucinglapar @kucinglaparssss]

@MrAhmadAwais Sure, with open source models, we can re training, and make some project with it. Your commandcode can help make some niche community apps, with some good result.

[Hassan @nutlope]

Ran 10 more tests comparing GLM 5.2 & Opus.

On average, GLM 5.2 produced 2x the tokens but was still faster + 3x cheaper with similar quality!

I'm open sourcing all these tests tomorrow, including the code, my prompts, and the token/cost stats.

[Kilo @kilocode]

It's live. Pick Auto Efficient in the model picker, or turn it on here: https://t.co/kinRtHqjoS

Matching the model to the task is the biggest thing you can do about your AI bill. This does it per request.

[kucinglapar @kucinglaparssss]

@MrAhmadAwais How about creating custom main agent? So user can add their custom agent, instead not just Build and Plan agent ? In github copilot and opencode, there are docs to create custom agent. And I have created some agent to our workflow
Maybe you can check it?
https://t.co/imfTepETEA