**Julian Goldie** (0:00)
If you can't use Fable 5, here's exactly what to use instead, and it's actually better. I'm gonna show you some side by side comparisons of exactly how this works. And let me just tell you what happened this week, because it's wild and it changes how you should think about your whole business. So on June the 9th, Anthropic released a model called Fable 5, the most powerful AI model that ever put out to the public. Three days later, just literally three days, the US government ordered them to pull it. So worldwide, it was gone for everybody. So if you went to use Fable 5 today, for example, you hit a wall, it's not there anymore. The best model on the market just vanished in 72 hours, for reasons that had nothing to do with it. Now, here's a part that actually matters. You can still reach that same level of intelligence today without Fable 5, and you can do it for about half the price. So there's a new model, it's called Fusion 5, and for about half the price, this can help you achieve Fable 5 level stuff. Now, this just came out from OpenRouter, and the time on this is almost too perfect, because Fusion proves the exact lesson, the Fable 5 takedown just taught everyone the hard way. I'm going to walk you through what got pulled, why it matters to you, even if you've never used Fable 5, and exactly how Fusion gets you that power back.
So stick with me, because the takeaway here will change how you set up your AI for good.
And first, let's talk about the facts. So Claude Fable 5 launched on June the 9th, 2026, Anthropic ordered the most capable model they'd ever released to the public. Had 1 million token context, strongest coding, research model out there. Then on June the 12th, that gets taken down. They cut off access to MiFaS 5 and also to Fable 5
And so Fusion actually helps us achieve Fable level performance on deep research tasks better than state of the art performance using panels. And it works by having a set of models working together. Then the judge fuses the answer and then you get one answer. And you have like multiple models working in parallel. So instead of leaning on one model, Fusion takes a panel of top models, usually three to five of them, and ask them all your question at the same time. So for example, in these tests, what we actually did is we used a combination of Gemini, Claude, Groq Build and Gemini 3.5 Flash. So these are the agents on the panel. And so if we have a look over here, we've got side by side comparisons of Opus 4.8 versus our panel of just Opus 4.8 of agents, right? So we've got the Fusion over here and we have Claude Opus 4.8. So which one performed the best? Well, let's open this up first of all. We can check this out side by side. So this is the option from Claude 4.8.
Now, one of the things I want to say is when you actually use this, it's super buggy. And what I mean by that is I literally can't move backwards.
Like I can't do anything, which is pretty interesting. I mean, it looks super nice, but it's pretty hard to navigate here. Let's try the other one. So this is the example that we got from the Fusion model.
Number one, it's much easier to move around and navigate. Number two, you got more details like the birds flying here. You have the reflection even of the glaciers on the lake, and you can move around and do stuff, right? So overall, this is a much nicer version of the original. I actually prefer the colors in the original, but I would say just in terms of navigating it without it being so buggy, I would say the Fusion won that. It would be interesting to know what people thought as well. Next up, we have a Minecraft style voxel world. So we've got Fusion over here, and we have Opus 4.8 over here. Let's try this out. So this is from Claude, and it says W A S D move. I literally can't move, right?
And the buttons don't actually work, which is kind of annoying. The only one that works is escape if I press escape. So it's super buggy, as you can see. You can't, oh, you can't move around now, but you see how that wasn't working for the first few minutes. It's pretty cool, though. I mean, that's, that is good. Like Opus 4.8 on its own, still the best model out there. Of course, it is a little bit buggy. Like for example, this. Now let's try out the other version. So this is the one from Fusion panel, right? So it works straight out of the box, seems a bit cleaner, less buggy when we actually use it. And you can see here that we can easily navigate it. Which one looks better? For sure, it's Fusion, right? So if we look, this is Claude 4.8. You see how the ground is not properly detailed here?
11 more minutes of transcript below
Try it now — copy, paste, done:
curl -H "x-api-key: pt_demo" \
https://spoken.md/transcripts/1000651996090
Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.
From $0.10 per transcript. No subscription. Credits never expire.
Using your own key:
curl -H "x-api-key: YOUR_KEY" \
https://spoken.md/transcripts/1000772920355