Vibe Check: GPT-5.4—OpenAI Is Back

GPT-5.4 is fast, opinionated, and good enough to tempt our Opus loyalist

by Dan Shipper and Katie Parrott

Was this newsletter forwarded to you? Sign up to get it in your inbox.

OpenAI released its latest model, GPT-5.4, today. After testing it for the past week across real engineering tasks, code reviews, and planning workflows, we can say that OpenAI is back in the coding race. Here is a taste of what we found:

GPT-5.4 won every planning test we threw at it, wrote code that's easier to review than its predecessor, and ran about twice as fast as Opus. Cora general manager Kieran Klaassen, our resident Claude Code loyalist, now reaches for GPT-5.4 daily.

But GPT-5.4 also tried to redesign an entire login system nobody asked it to touch and told us tasks were done when they weren't. The model's ambition is both its best feature and its biggest liability.

Read the full Vibe Check for the complete analysis—head-to-head comparisons against Opus 4.6, GPT-5.3 Codex, and Gemini 3.1 Pro, plus Kieran's LFG benchmark results across seven real-world coding challenges.

Read the full Vibe Check

We're also hosting a livestream about the release on YouTube at 1 p.m. ET.

To read more essays like this, subscribe to Every, and follow us on X at @every and on LinkedIn.

We build AI tools for readers like you. Write brilliantly with Spiral. Organize files automatically with Sparkle. Deliver yourself from email with Cora. Dictate effortlessly with Monologue.

We also do AI training, adoption, and innovation for companies. Work with us to bring AI into your organization.

Discover Every's upcoming workshops and camps, and access recordings from past events.

For sponsorship opportunities, reach out to sponsorships@every.to.