Remember just a few days ago, everyone was predicting the release of Sonnet 5? Well that aged beautifully considering that Anthropic just dropped Opus 4.6. Gee, I sure hope we don’t get a new model in like, 15 minutes—Oh wait. What’s happening at OpenAI?
In today’s episode of ‘A new model just dropped’, we actually got 2. Anthropic launched Opus 4.6, the direct successor to 4.5, and without missing a beat OpenAI delivered 5.3 codex.
One thing both mathematicians and vibe-coders can agree on today is that
5.3 > 4.6 > 3. From what i’ve seen, there is a massive leap in quality from 5.2 to 5.3.
I have yet to fully test both models—though it seems 5.3 is absolutely insane. I used the model to redesign my portfolio. It is safe to say the model has a colossal improvement in UI design.
I’ve also noticed the model is incrementally faster then its predecessor—as well as being more conversational whilst working. Previously with 5.2, giving it a task would result the model directly tool calling. In contrast, 5.3 clearly states what it plans to do before doing it. If 5.2 approached a task with the wrong direction, you would be unaware till it completes, as opposed to 5.3 which keeps you posted the entire time.
Well how about the other model that dropped today—4.6? It isn’t exactly the best… in fact it might not be a stretch to say it’s a downgrade from Opus 4.5.
So what about Google
A few months ago, I decided that it was a great time to invest in Google. Fast forward to today with OpenAI and Anthropic having increased more than I ever thought possible.
Hopefully the next Gemini model changes things because it isn’t looking good for us right now. Even at this point, I strictly refuse to sell because I still have a little faith left. It’s more than an investment strategy—iit’sa coping mechanism.
Maybe in 3 or 4 weeks we will look back on this and laugh at how bad the OpenAI and Anthropic models compare to Google’s. For now though, we wait.
Closing thoughts
We got a great model from OpenAI today and a decent-ish model from Anthropic while Google remains silent. However, if you weren’t already aware; with the release of the new Codex app, ChatGPT Plus and Pro users get a temporary double rate limit in app. Free and Go users can also try it while it lasts.
Anthropic also put out free $50 USD in credits for Claude Code, which you can claim under settings → usage.