codex is now more efficient

@ashenbot

8.6K followers · usually 1.3K views a post

Views

13.3K

Likes

663

Comments

58

Shares

64

Selling · SaaS

Codexseen in 40 videos →

Why it worked

The creator leverages a 'insider' tone by referencing a specific developer (Tibo) and his habits, which builds community trust and makes the technical update feel like a personal win for the users.

Summary

The creator discusses recent efficiency updates to the AI tool Codex that have resulted in lower serving costs and better performance. He highlights the developer's frequent, generous resets of usage limits as a key reason for his preference for the platform.

Structure

1Announcing efficiency gains for Codex
2Explaining the developer's habit of resetting usage limits
3Detailing the specific performance improvements
4Connecting the technical updates to better user experience

Product placement

The video is entirely focused on the AI tool Codex, using a screenshot of a tweet from the company as a visual aid.

On-screen text

codex is now more efficient After deployment, we applied GPT-5.6 Sol to advance the frontier of efficiency by making itself more efficient to run. The results: - 20% lower serving costs from production GPU kernel improvements. - 15%+ better token-generation efficiency from improved speculative decoding. These optimizations across the stack compound to unlock the most performant models at every point in the cost-intelligence curve.

Original caption

codex limits > all #ai #technology #techtok #codex #chatgpt

More from @ashenbot

Their other posts in the index, biggest breakout first.

Transcript

After deployment, we applied GPT-5.6 Sol to advance the frontier of efficiency by making itself more efficient to run. The results: 20% lower serving costs from production GPU kernel improvements. 15%+ better token-generation efficiency from improved speculative decoding. Codex enjoyers, rise up. Let it be known. Just yesterday, last night, Tibo, the guy who works at Codex, reset all usage limits. For those of you that don't know, on Twitter, this guy Tibo works at Codex. Every other day, it seems like once a week minimum, he'll just be like, by the way guys, reset all your limits. So if you're ever on Codex and you're like, wait, wasn't I at 20%? But then you open it back up and you're at 100%, it's because one of those guys literally just presses a button. But not just that, six hours ago, they also tweeted from the actual account that they made it more efficient to use 5.6 Sol today. Without getting too into it, all you really need to know is that they now have 15%+ better token-generation efficiency and 20% lower serving costs from production GPU, meaning if you're using Codex, you will most likely experience higher usage limits because whatever they're doing in the back end is more efficient now. Which is one of the main reasons I love Codex. They have really, really generous usage limits. And if you ever experience one of these random resets, it's now in addition to more generic like efficient and productive stuff happening in the background.