I Tested Opus 5.5 vs Fable 5.1 on 7 Real Use Cases (Not Even Close)

0:00Claude just released a new Opus 5.5 model and according to the benchmarks, it's even better than Fable 5.1 and even cheaper than Claude Opus 5. But we all know that these benchmarks are not that reliable. So for this video, I tested it across seven real use cases that apply to my day-to-day like slide decks, research, data analytics, website design, and automating work. And for each one, I'll break down which model actually did better, how long it took, what it cost, uh so you can know which model to use for what use case. Starting with the first one, creating slide decks. So I gave Fable 5.1 and Opus 5.5

0:34the exact same prompt, um to help me prepare a slide deck with a breakdown of my marketing performance over the last 30 days compared to the previous month. So it had to go through some of my connectors to get the data. I also asked it to add in charts, graphs, or visuals whenever it can. And I gave it some reference images of a slide deck to see how well they can follow uh a specific style. Now Fable 5.1 gave me this slide deck. As you can see, it followed the style very well, but almost all of the graphs and tables like here and here uh are sort of messed up. Either it's

1:04really small or it just it doesn't show it clearly. There just seems to be a lot of sort of issues with the with the formatting, as you can see here. Also, I wasn't a huge fan of these copies here. It's sort of pretty hard to read and pretty verbose. So I gave Opus 5.5 the exact same prompt and it gave me this one. Again, it followed the style very well, even put a little bit of a

1:23gradient in here. Um in general, it just looks a lot cleaner. Um you can see it actually did the bars, the graphs, the charts uh actually really well. Although there's also a slight mistake here in the formatting, uh but in general, it just looks a lot cleaner, and to the point. I also like the copy a lot more. It's a lot more to the point, uh a lot clearer, and less verbose, as you can see here, which is something Entropic also specifically mentioned that Opus 5.5 uh is better at writing copy. Now Opus 5.5 took 22 minutes to build this, Fable 5.1 uh 21 minutes. Opus actually

1:56used 30 million uh tokens instead of 20 million versus Favon, it means it probably did more in-depth research on my marketing analytics before creating the presentation, but it was still half the cost of what Favon cost me, so very clear winner for this use case. Now, for the second use case, I wanted to try one more use case for the design capability, so I gave them both the same prompt to redesign the landing page of a new app I recently launched, which is a second brain app called Balda. As you can see, it's a pretty basic website at the moment. In the prompt, of course, I gave

2:26it the original website, and I also added in this reference link of a website that I liked to see how well it can follow a style from another website. So, I turned this website link to follow the style from, and I also added in a GitHub repository with a liquid glass effect, so it can apply that to some of the elements. This is from from Apple, if you know it. And lastly, I instructed it to add in some scroll animations using Hicksfield. Favon 5.1 gave me this output. It did follow the style pretty well, although

2:55the layout changed significantly. It also applied this liquid glass effect, as you can see here at the menu. It added in these scroll animations, as you can see it see. It shows some screenshots of the app, which is good. Again, we have more scroll animations here. It looks all right. Also, the copy is decent, pretty good. It also added in some of these hover animations from the original website, which is good. Yeah, more scroll animations. I'd say overall, pretty good, but it doesn't blow me away. Then, Opus 5.5 gave me this output with the exact same prompt. First of all, it looks a lot more like the

3:30original reference website that I gave it. Again, it applied the liquid glass, and the scroll animation here looks a lot smoother than the previous one with this sort of 3D effect. I'd say the overall layout looks a bit cleaner. The scroll animations look a little bit more smooth. The copy seems to be more to the point, too. It even added in this

3:50table comparing it with other solutions. It added it this in from the reference too. It also has these hover animations. So, honestly, looks very good. They even added in a FAQ, and here at the end, you can see this is a nice little effect, too. So, in my opinion, again, Opus 5.5 clear winner in the output quality. Now, Opus took 17 minutes, while Fable took 14 minutes. Fable used 5 million tokens, Opus 13 million, but again, Opus is just far cheaper, almost half the price as Fable. So, for design use cases, it's pretty obvious you want to go with Opus 5.5. So, for the third use case, I wanted to test the research

4:24capabilities of these models. So, again, I gave them the same prompt. In this case, I wanted to do a competitor research on a new video topic that I'm doing to identify the best titles and thumbnails for the new video I'm preparing. So, it basically had to go through the last 7 days of videos on YouTube around this topic to identify a

4:44good title or thumbnail for this video. Now, there's actually a big difference in output here. First of all, Fable 5.1 here analyzed 50 videos on YouTube around this topic in total. It did rank them by performance very well, and it gave me at the end a breakdown of some thumbnail outliers, what they call. Now, Opus 5.5 gave me a very similar output here, but Opus actually analyzed 168 videos, and even analyzed 59 non-English videos, which is quite creative. Again, gave me a very similar title with videos ranked by

5:18performance as this cool animation, too. And at the end, it also breaks down some of the thumbnails. So, similar output, but Opus just went far deeper in its research. Now, Opus took 15 minutes for this task, Fable took 14. Opus, again, spent far more tokens here, more than double Fable's, and therefore even costs slightly more than Fable, but because it did just a lot more in-depth research, I'd still say that Opus just wins overall for this one. Now, I wanted to to one more research use case to see if it still performs better on a different type of research task. So, I also tested it on helping me prepare a pre-outline

5:53for a YouTube video on strategies and tips for improving AI adoption across a business by going through and researching through Claude's big library of customer case studies to find good tips, strategies, and insights around how some of these companies improve AI adoption. Now, Fable 5.1 gave me this output. It made a table here with the 13 different strategies and tips that I

6:16found. For each one, it also broke it down by a one-liner. It gave me the proof with a real quote where it got it from, together with the link. The tips themselves I found pretty good and not too obvious. So, overall a very good output from Fable. Now, Opus 5.5 did it slightly different. It put the 15 strategies into four buckets. So, it maybe put a little bit more structure already into the video. It again gave specific proof where it got it from, together with the

6:45link. Now, the strategies themselves are also pretty good. Start with the task people. Hey, there's some very similar ones that Fable found. So, overall very comparable and similar output. Both of them are really good. Now, Opus 5.5 did this in 13 minutes, while Fable took 20 minutes. Now, you can see a huge difference here in the tokens used because Opus 5.5 actually spun up lots of sub agents with another model, Sonnet, and that's why it's still actually half the price even though it used 5 million

7:15tokens, mostly with these sub agents. While Fable Fable mostly did it on himself by himself, and that's why it's $15. So again, based on cost and speed, I'd still say here Opus 5.5 wins again. So, definitely recommend using Opus 5.5 for your research task. Now, very quickly before continuing, if you enjoy this type of content, you'll really enjoy my AI Accelerator community where I just launched three new beginner to advanced Claude courses that walk you through step-by-step how to actually use Claude to automate work as a non-technical professional or business owner. We also have unlimited one-on-one live tech help to help you with any problems you might face. We have a full

7:50courses on building your own AI business and getting AI clients fast and a big library of skills and resources across all business departments that you can use and customize for yourself and a community of professionals, business owners, and AI agency owners actively using AI on a day-to-day. So, if you want to dive deeper into this, definitely check it out in the link in the description below. So, the next use case I tried it on is creating documents. So, in this case I asked it to turn one of my YouTube videos into a written lesson inside of a document so that I can add it into my accelerator

8:19community as a as a sort of a lesson. So, I asked it to create an explainer doc on a video that I did around prompting. Of course, fed it the YouTube link and I also gave it access to the /watch scale. So, Claude can actually take screenshots from the video and get the second-by-second timestamp transcription. Now, Fable 5.1 then gave me this Google Doc here. Uh, it did apply my brand colors and it also took screenshots from the video as you can see here. Uh, now overall, it looks decent. The layout is not perfect, but it does look good and you can see it even took screenshots

8:53and then highlighted specific parts and added text to explain what's important here. So, overall, uh, pretty impressive here. It took screenshots from my prompt to explain things. So, pretty impressive, but the layout is not perfect. Uh, but overall, I would say this is this is a very good output. Then Opus 5.5 gave me this. Again, applied my brand style uh, very well, but you can also see right away that it's better formatted. I also like the copies directly a lot more. They're a lot more clear and straight to the point, which again is something that Opus 5.5 is specifically trained on supposedly, uh, but I do see the difference. And I can

9:31see it also took uh, screenshots from the video. It also highlighted the important parts and even added in an arrow with an extra explanation. Uh so, this is actually uh really impressive what it can do with a one-shot prompt like this. So, you can see it highlights everything with an arrow. So, this this boxes are perfectly aligned in this

9:50case. So, again, I would say both are pretty impressive, but a clear winner on the output Opus 5.5. Now, Opus took 14 minutes, Fable took 19 minutes. Opus used 13 million tokens, while Fable used 9 million. Uh and it was basically half the cost, $5. So, again, Opus wins for creating documents. Then I wanted to try it on a data analysis use case, which I was honestly very impressed by with Fable 5.1 to see if Opus also outperforms for these types of tasks. So, again, I gave it the same prompt, in this case to create me a customer intelligence report based on all our customer support calls, Q&As,

10:24and uh customer support tickets over the last 2 months. And then based on all the problems it identified, give me a report uh with video ideas based on uh the common problems, a list of potential ways to upgrade or improve my product based on customer problems, and thirdly, a comprehensive breakdown of my customer profile. Fable went off for more than an hour and analyzed more than 500 customer support calls, 15 Q&As, and 266 customer uh support tickets. It then gave me a a brief summary first of the top problems and top things it identified. Then it gave me 14 video ideas based on uh the most common problems that my members

11:02faced. And for each of these videos, it also backs it up with real quotes from these calls on why this these are uh why this is a good idea. So, it did that for all these 14 video ideas. Then it also gave me a prioritized list of potential things to improve in my product, again backed up by real quotes from these uh customer support calls. And lastly, a very comprehensive breakdown here of uh my ICP or customer profile, uh which is honestly uh very good and useful. So, honestly very impressive and useful report. Now, Opus 5.5 gave me this and the first thing I noticed is that it

11:39only analyzed 248 customer support calls instead of uh 555 for Fable. It gave me a very similar layout. First the summary with the top problems it found, 14 video ideas uh just like Fable, but in this case it didn't actually back it up with direct quotes. It did give me links to the specific um meeting transcripts, but not the direct quotes. It gave me 13 potential ideas for improving my product, uh but again it didn't actually add in references or quotes uh where it based this on. And I would also have to say that the Fable ideas were better than this one. This is probably because

12:16it analyzed more. And it also gave me a customer profile breakdown, which is also good, but again I would say Fable was a lot better at this one. Now, and the reason is also because Fable spent an hour and 15 minutes here while Opus spent 40 minutes. And it also spent 33 million tokens versus 7 million tokens. Uh and this report cost me $100 versus

12:35$62. Uh now, you can clearly see that maybe for these data analytics uh use cases Fable just put in a lot more effort and therefore we get a better report. Now, I don't know exactly why this is, but this is a much more complex and context-heavy uh sort of task. And what I did notice here and why this token spent so much higher is that Fable span up many more sub agents than Opus. And Fable is supposedly a very

13:00good orchestrator model. Um so, it might mean that for tasks where it has to go through a lot of data or a lot of context to get to an output, Fable just spins up more sub agents, puts in more effort to get to a better output. But of course, it is going to be more costly. So, you might still want to consider using Fable if you don't mind the cost of course for these extremely complex or data-heavy tasks. So, for the last one I wanted to try one more use case to confirm that uh Fable just goes through a lot more context than Opus by using these models together with my

13:32second brain. Also, to be able to see how much context from the second brain it would actually read before giving me an output. So, I asked it to go through my second brain uh to do an analysis on the trajectory of my business over the last year to get a better overview of the progress we made, what decisions we've made, what we've actually stuck

13:49to, and what bottlenecks we've resolved. Now, I can't show everything here because there is some sensitive data, but Fable again went through a 199,000 in my second brain and broke the progress down first here into phases over the year, then a a table with decisions we've made, what we actually stuck to, uh what we what changed in our decisions, an overview of where we actually made progress, an overview of everything that got got fixed, and everything that's still open for the rest of the year, and an overview here in a table of what actually moved the needle. So, really useful overview, and it did take a lot of context from my

14:21second brain into account. Now, Opus it is a lot faster and only read around 50 documents. So, four times less than Fable to get to this. It put it in a similar structure, laid it out in phases with a table with the decisions we've made, what we stuck to, etc. Um what actually got fixed, uh but it's a lot shorter, as you can see. It didn't add in anything on move what moved the needle or any conclusions. It's also just a lot less in-depth. So, again, a similar pattern for this one as the previous one, uh and basically what I'm seeing is when there's a lot of context

14:52involved, it seems Fable goes deeper, but of course, there is a significant difference in the cost. Now, Opus took 5 minutes here, Fable took 17 minutes. Um it used 3 million tokens versus 5 million tokens, and Fable cost me $17 and $2. So, again, similar pattern as the previous one. When there's a lot of context involved, it seems Fable goes uh deeper and reads just more, but of course, there is a significant difference in cost. So, in summary, both these models are very powerful models, but as you can see, Opus 5.5 wins across most use cases on quality, cost, and speed, especially for copy and design. I

15:26was really impressed. In In my opinion, Opus 5.5 should be used uh for almost all use cases. Now, there might be a slight argument to make that Fable 5.1, if you're willing to put in the extra spend uh for extremely complex tasks that require reading through lots of context or data, might still be worth using it, maybe because Fable is still a better orchestration model. Uh but, of course, you have to be fine with paying the extra cost. Now, there's one more note to make with this, which is a theory that I recently heard and I think makes a lot of sense, which is that Anthropic, behind the scenes, puts the

15:58effort levels of these new models far higher when they're just released. So, they sort of catch the hype and they look extremely impressive at the beginning, but of course, it does cost a lot of compute. So, after a couple of weeks, they turn down that effort level or reasoning to save on compute. And now, this is something that I studied actually around Fable, and I've personally sort of seen it, too. When it first launched, it went a lot more

16:21in-depth than it seems to do right now. Uh so, it's one thing to keep in mind, because you'll probably see Opus 5.5 uh performance declining a little bit over the upcoming weeks, but I'd still say Opus 5.5 is the model you want to use for most tasks. Now, if you want to dive deeper into AI, definitely check out my AI Accelerator, where I just launched three new beginner to advanced Claude courses that teach you exactly how to use Claude to actually automate work as a non-technical professional or business owner. We also have full courses on launching your own AI business or getting clients fast. We have unlimited

16:52one-on-one live tech help to help you with any problems you might face. We have a big library of resources, skills uh that you can use and customize for yourself, and a community of professionals, business owners, and AI agency owners actively using AI. So, definitely check it out in the link in the description below. And if you want to learn more about working with Claude,

17:10you can check out the video here above.