Korean

[Tibo @thsottiaux]

If you aren't yet bold enough to install the Codex app, you can stay in the presence of your orange crab and point it at GPT 5.6 Sol. Takes 5 minutes. Kudos to Theo for explaining one of the ways to get this done.

Step 1: Install CLIProxyAPI
Step 2: Connect
Step 3: Define following alias and enjoy claudex

```
alias claudex='CLAUDE_CODE_SUBAGENT_MODEL=gpt-5.6-sol \
CLAUDE_CODE_ALWAYS_ENABLE_EFFORT=1 \
CLAUDE_CODE_MAX_TOOL_USE_CONCURRENCY=3 \
ENABLE_TOOL_SEARCH=false \
claude --model gpt-5.6-sol'
```

If this gets blocked, I owe you a reset.

[Korean @Brody_iOS]

억만장자글 키배 또 시작이네 ㅋㅋㅋㅋㅋ

[Sam Altman @sama]

there are a lot of benchmarks that suggest 5.6 sol is the best model in the world right now, but the most reliable way to tell is that elon is obsessed with me again

[Yun-Ta Tsai @yunta_tsai]

Even Zuckerberg has to advertise his AI model on X.

X is the real arena.

[Korean @Brody_iOS]

@skcd42 I’m in Korea, actually. I just woke up and checked — it’s working for me now!

[Korean @Brody_iOS]

@ericzakariasson ⁉️ * Grok 4.5 has an advantage on CursorBench: an earlier snapshot of the Cursor codebase was accidentally included in training. The exact score impact is unclear. That data has been removed for future models, and was not present for Composer models in the past.

[ayv4zyan @ayv4zyan]

Plot twist: xAI just nerfed limits for paid Grok/SuperGrok users by switching to weekly quotas because demand is already eating the cluster. Even Elon doesn’t have enough compute for his own paying subscribers.

[Korean @Brody_iOS]

@gimhyeo02389130 herdr 👍👍

[Morgan @morganlinton]

I think a lot of people are using Fable High and Extra, when they can probably do most of their low and medium, and even some higher complexity work with Low and Medium Effort.

We have to realign how we think about models and the level of thinking depth we need from them now that we have something like Fable in the wild.

I will be doing more benchmarks, but the first one I ran shows Fable is actually cheaper than GLM 5.2 or Sonnet, and with higher accuracy.

[Claude @claudeai]

Sonnet 5 is a substantial improvement over Sonnet 4.6 on reasoning, tool use, coding, and knowledge work.

Its performance is close to Opus 4.8, at lower prices.

[David @dayonefoundry]

18% sounds like a lot already but the upgrade cost of RAM went up 100% from +$800 to +$1600

[Korean @Brody_iOS]

@OpenAI 설렜네 ;;

[Korean @Brody_iOS]

@XFreeze composer 2.5 👍👍

[Korean @Brody_iOS]

@grok Do we need to setup plan with "/plan" before using goal?

[Grok @grok]

Introducing /goal in Grok Build.

Execute long-running tasks autonomously, with multiple rounds of subagents implementing and verifying a single goal.

[JNS @_devJNS]

US banned Fable 5

Japan:

[Korean @Brody_iOS]

@LuminaXspace maybe july

[Tech Dev Notes @techdevnotes]

Watch out Grok 1.5T model is coming soon

[Grok @grok]

Manage multiple agents at once in the Agent Dashboard.

See what each is doing, reply to the ones that need you, and dispatch new tasks.

Try /dashboard in Grok Build.

[Korean @Brody_iOS]

@skcd42 @aether_oracle Tools like abtop and tokscale show how useful a clean TUI dashboard is for token usage and session stats. A similar dedicated view in Grok Build would be really valuable — especially for checking cache hit rates.