Ale Bencz

[Ale Bencz @AleBencz]

@MonoGameTeam There was no reason to keep this code just for myself... especially since it's a study project... I already have several open-source projects... this is just one more :)

[Ale Bencz @AleBencz]

Etervale is now open source!

Started around 2010 in XNA, later moved to MonoGame, it's an incomplete Priston Tale reimplementation in C# using the original assets.

https://t.co/vqCrqD6dyA

#OpenSource #GameDev #PristonTale #MonoGame

[Ale Bencz @AleBencz]

@RonaldoGogoni @hsramos isso é ahahah
não falei q eram baratos, apenas falei as alternativas xD
qq coisa q tenha memoria ram hoje em dia, ta 800% mais caro

[Ale Bencz @AleBencz]

@hsramos @RonaldoGogoni tem varios... Radxa, OrangePI, BananaPI .... e por ai vai

[Ale Bencz @AleBencz]

@buildwithrajath Well… that is it…. Opus 5.5 nerfed xD

[Ale Bencz @AleBencz]

@oviniciuslana

[Ale Bencz @AleBencz]

@thsottiaux

[Ale Bencz @AleBencz]

@thsottiaux Shameful... TRULY, TRULY SHAMEFUL!!

[Ale Bencz @AleBencz]

Well, it’s pretty clear OpenAI has a massive technical quality problem. They can’t figure out how to deal with hardware shortages by actually optimizing their own models (just look at what Anthropic did...), so they make users pay the price instead

#openai #codex #chatgpt

[Ale Bencz @AleBencz]

The state of the art right now, between OpenAI and Anthropic

#openai #codex #claude #claudecode @thsottiaux

[Ale Bencz @AleBencz]

@steipete These assets are from Outlive

[Ale Bencz @AleBencz]

@pcshipp

[Ale Bencz @AleBencz]

@rusti_red @DanDr1s Maybe less… let’s check in 2 or 3 days

[Ale Bencz @AleBencz]

GPT-6 Sol is an absolute joke

#openai #chatgpt #codex @thsottiaux

[Ale Bencz @AleBencz]

@ramirezfdes2022 Leia isso: https://t.co/Sz8LY5VQxB

[Ale Bencz @AleBencz]

@giacomogh96 @thsottiaux I’ve seen poor-quality TypeScript code too, on top of it not giving a shit about AGENTS.md

[Ale Bencz @AleBencz]

@giacomogh96 @thsottiaux Another thing I noticed that GPT-5.6 Sol doesn’t do: it got stuck on a basic implementation and kept looping, even after resetting the context. GPT-5.6 solved it with an elegant, high-quality implementation