Switched to GPT-5.6 Sol inside Codex. Refactoring a Hexagonal Architecture backend (DUB Engine) — Celery queues, ports & adapters, multi-service DI — and it holds project context across long agentic runs. No re-explaining the architecture every turn. That's the real unlock.
Or… what if we gave you $100 in Codex credits if you tell us what you love about GPT-5.6 Sol or why you switched?
Tweet it, claim your gift, enjoy more usage. First 10k get the free tokens!
https://t.co/8mU93eA13i
@ersinkoc Promt yazarken codex gibi agentlarımı skillerimi anında @ ile ya da / ile çağırabilmeliyim bence. Desktop kullanıyorum şu an. Güzel bir kapsamlı tanıtım videosu şart olmuş. Bunun öğrenme eğrisi çok yüksek.
@ersinkoc Promt yazarken kendi kendine gönderdi bu bir özellik mi? Prompt iyileştirme açıktı kapattım. Ama bir bug kalmış olabilir. Yazarken direkt göndermiş. Bir de @ ile dosya seçtirmiyor. Mesela backendi komple etiketlemek istedim yapamadım. Elinize sağlık.
@thsottiaux He used to eat from the weekly limit until the mission was over. Now the task is interrupted. This feature of the new Codex is very bad. I also feel like limits are being cut very very much. Even moderately, it ends up on a few tasks. 5.5 was better at this.
@thsottiaux "frontend." "When I try to do development with With Three.js it's pretty bad. I tested it with all the plugins and skills. But the results are like GPT 3.5.