Zorbey Torunoglu

i love building things

[Zorbey Torunoglu @torunogluzorbey]

@fanofaliens how long did it take?

[Zorbey Torunoglu @torunogluzorbey]

@hakmgpt don't buy access, buy usage. subs won't last, api will.

[Zorbey Torunoglu @torunogluzorbey]

@melvindvivas an engineering workflow from someone who actually pays for their tokens is worth more than ten from someone who doesn't (poteto). would love to see

[Zorbey Torunoglu @torunogluzorbey]

@elshayib_ even deepseek's flash is way ahead and many times cheaper. there is zero reason to use grok at this point unless you are elon

[Zorbey Torunoglu @torunogluzorbey]

it is so sad to see grok's flagship model fall behind deepseek's flash in every way

[Rafa @RafaCrackYT]

🇫🇷 I asked 4 AI models to take me back to Paris in 1996. 100% code. 🇫🇷

Same prompt, one-shot:
Opus 5.5: $73.80 · 2h56m
Gemini 4 Argon: ~$2.50 · 30m
DeepSeek Flash 1017: $0.34 · 46m
Grok 4.7: ~$7 · 2h07m

Opus nailed the vibe, but DeepSeek got almost on its level for 217x less 🤯

[Zorbey Torunoglu @torunogluzorbey]

i block anyone who starts a post with

🚨 codex users!!!!
🚨 claude users!!!

[Zorbey Torunoglu @torunogluzorbey]

@FishAudio DRAMA

[Zorbey Torunoglu @torunogluzorbey]

@rezoundous many seek stability rather than the best model. after all those sudden 50% cut, regular model at capacity errors; don't think so.

[Zorbey Torunoglu @torunogluzorbey]

@TimJayas benchmarks are marketing in a lab coat. totally normal, just don't vote with your wallet yet

[Zorbey Torunoglu @torunogluzorbey]

@buildwithrajath there is no 'fix' coming. limits and quota burn aren't bugs, they're business model. it'll always swing up and down. don't depend on one tool.

[Zorbey Torunoglu @torunogluzorbey]

@JamesMalsawm the $3.99 you save goes straight into fixing what it wrote

[Zorbey Torunoglu @torunogluzorbey]

@fahnx8 slow but so token efficient. it saves tokens by barely producing any 😂

[Zorbey Torunoglu @torunogluzorbey]

@melvindvivas mind sharing more about how you use your bots?

[Zorbey Torunoglu @torunogluzorbey]

@_jayyy18_13k gemini 4 argon about the be the best model i never use

[Zorbey Torunoglu @torunogluzorbey]

@NathanJBarnes @elshayib_ makes sense!

[Zorbey Torunoglu @torunogluzorbey]

@elshayib_ oh yes you are totally right about that, they are unstable. I'll dump oai and get the new X ultra sub for that very reason.

[Zorbey Torunoglu @torunogluzorbey]

@bukovinafilip bit harsh. grok handles most engineering work fine if you know how to drive it, and near unlimited usage is a big deal.

[Zorbey Torunoglu @torunogluzorbey]

@NathanJBarnes @elshayib_ how do you prefer to let your bots use codex?

[Zorbey Torunoglu @torunogluzorbey]

@worldofaiprompt not just before launches. gemini needs more time in the oven, full stop.