Transcription
So, Anthropic has just released Opus 5, and I have to admit the benchmarks look pretty incredible. They're obviously positioning this as an alternative to Fable 5, but the question is, does it actually change how you use AI every single day? And most importantly, does it change any of your workflows automations that you currently already use? And the truth is, after experimenting with it, I don't think it's going to change the needle for most people.
So, if we look at some of the benchmarks here, agentic terminal coding, knowledge work, novel problem-solving, again, that is actually a really agentic search. It's pretty much beating Fable on almost every single category. And the ones that it's not beating it on are really like maybe one or two percent differences. So, the natural reaction at this point is to think, okay, well, why do we still need Fable at all? I mean, this is just a much better model than anything we've seen before. I mean, you even have something like the business workflows, which is kind of a newer category, but I think for the layman and for people who are doing it for kind of automations in your business, which I think is the most practical use case, um it scores at 26%, which is a lot higher than any other model before. That part particularly, once that starts to get into the 70 and 80% mark, I think that's really when you're going to start to see the I don't know if you want to say the singularity, but it's going to become so much in the public psyche, guys, that everybody's going to be using AI. That will really be the key barometer right there.
And here's the thing, right? Tariq from Anthropic himself said that Fable 5 is still a better overall generalist model. So, what does that mean? For the stuff that you're already using Fable 4, reviewing, planning, brainstorming ideas, which is really where Fable shines, I don't think any of that is going to change. You're still going to be using Fable. I think ultimately the biggest thing that is going to change is it's going to be the decision, do you want to use something like Sol 5.6 as your daily driver, or do you maybe use Opus 5 as your daily driver now? I know for me personally, I don't think this is going to change my workflow. I've already been using Fable to brainstorm and review all my ideas. I think most people have. There's not that many people that are going to be using Fable to do the actual coding just because it's so token intensive and you have such a little usage, especially on the plans that you have. The Codex experience is just so much better than the Claude code experience. Yes, they all both have computer use the cloud plugins, files and terminal, but when I'm using Codex, the browser workflow is a lot better. The computer use is literally second to none. I mean, what they did with computer use is just absolutely incredible. The plugin experience is much simpler and the recent release of their new voice mode is just so much better than what Claude has right now. So, the truth is that the model is only half the decision. Ultimately, the harness itself is still going to be better with Codex and because of that, I don't think my daily workflow is going to be changing that much.
So, then I guess the better question is who does benefit from this new release? And so, if you're looking at it here, it's really going to be developers, the ones that are doing coding all day long, autonomous coding, debugging code of view, working at really large code bases. So, you can even think of like developers that are working at major enterprises like Microsoft or Google and all that, assuming that they're not using their own models. That's probably going to move the needle the most for them, especially cuz they're already worried about API usage and not burning a ton of tokens. This is going to be a real meaningful upgrade for them.
In terms of business operators, like myself or like a lot of common people who run small businesses or medium-sized businesses, this is really not going to change that much. Yes, automation bench improved. So, some of the automations that you run may work a little bit better, but the truth is most automations, if you scope them out well enough, tend to work really well anyway. And so, it was already really strong at that. So, any type of real-world advantage is not really guaranteed with the release of Opus 5 now. And if we're really being honest, it's not the model that's usually the issue for anything that you have going on in your business. Most of the issues that you're coming across are still going to remain the same. Frontier models have more than enough intelligence now. What they're usually missing is context, weak workflows, uh not having the right tool access or unclear instructions. And guess what? That is the exact same thing as when you're working with a real employee. Fix the context, fix the So, fix the context, fix the workflow, and fix the tools before chasing another model or constantly hopping back and forth.
For me personally, I've already been using Fable primarily as a planner and reviewer, so none of this is going to change. The only difference is maybe I won't delegate out to solve 5.6 to do the actual work. But realistically, with the workflow that I have, I don't even think I'll be changing much of this. If anything, I'm just going to use Opus 5 to be to review in more and more checkpoints than just having Fable, right? So, maybe I'll have Fable review much larger checkpoints, and maybe I'll have Opus just review along the way on a more granular level. So, long story short, if you've already been using Fable and you delegate a bunch of your actual work to solve 5.6 as your driver, keep doing that, nothing's going to change. If you only have a Claude code plan, obviously you can use Opus 5 as your driver now. Believe this channel is also to kind of cut through the noise and see what's actually going to make you more productive on a day-to-day basis. So, hopefully you enjoyed this quick video. I enjoyed making it. If you have any other questions or concerns, anything you'd like to see, just shoot me a comment and I'll see what I can do. Thank you all.