| Highlights: |
|
If you use Claude every day, the model menu has become a little crowded.
In early September, Anthropic released Claude Fable 5.1, its highest-capability model for difficult coding, research, and knowledge work. Now it has added Opus 5.5 and Sonnet 5.5, two models aimed at a much broader range of everyday work.
That raises a reasonable question: if all three models are good at documents, research, analysis, coding, and other professional tasks, which one are you supposed to pick? The differences are becoming smaller, but there are still some useful distinctions.
Sonnet 5.5: The Everyday Option Got Much Stronger
Anthropic released Sonnet 5.5 on September 28. It’s designed for well-defined work such as analyzing information, fixing code, building presentations, working with spreadsheets, and creating polished documents.
Compared with Sonnet 5, Anthropic says the new model generates responses more than 30% faster and can cost up to 30% less per task because it generally needs less processing to reach the answer.
For a regular Claude user, speed may be one of the easiest differences to notice. Suppose you ask Claude to create a presentation from a report. You review it, ask for a different structure, replace two slides, add another chart, and then tighten the wording. Faster responses matter much more across that entire back-and-forth than they do on one isolated prompt.
Anthropic is also paying particular attention to the quality of the finished files Claude creates. In one of its internal tests, Sonnet 5.5 received a company's earnings materials, an earnings-call transcript, and a slide template and was asked to create a 10-slide operating review. Two reviewers considered the first draft ready to send without changes.
That is Anthropic's own test, so it shouldn’t be read as a promise that every presentation will come out perfectly. However, it does give a good indication of what Sonnet 5.5 is being built for: getting normal work done quickly and producing something that requires less cleanup afterward.
Opus 5.5: For Work That Requires More Reasoning
Opus 5.5 arrived a few days earlier, on September 22. This model is intended for work that involves more uncertainty, more steps, or more judgment. Anthropic points to areas such as research, financial analysis, business workflows, computer use, and complicated coding tasks.
Think about the difference between asking Claude to summarize a quarterly report and asking it to investigate a company's performance using several sources, reconcile numbers that don’t immediately match, identify the important issues, and turn the findings into a recommendation. The second task gives Claude much more to work out on its own. That’s where Opus 5.5 is supposed to have an advantage.
In one Anthropic research test, Opus 5.5 had to find a difficult-to-locate earnings release and use it to prepare a report. Sixteen of 18 attempts met Anthropic's quality threshold, which required every figure and quotation to be supported by the source. Fable 5.1 and the previous Opus 5 did not pass the same threshold in that test.
Anthropic also tested Opus 5.5 on a fictional merger analysis. Claude had to build a financial model in Excel and then create an executive presentation from it. Opus 5.5 completed the work in 63 minutes, compared with 93 minutes for Opus 5, while producing a more detailed model and a presentation Anthropic considered easier to read.
For someone using Claude at work, Opus 5.5 makes the most sense when you’re giving Claude a problem rather than simply a task.
What's the Difference Between Sonnet 5.5, Opus 5.5, and Fable 5.1?
A few years ago, choosing between model tiers often meant making an obvious tradeoff between speed and intelligence. The gaps are much narrower now.
Sonnet 5.5
Sonnet 5.5 is the model I would think of first for normal, well-defined work. You know what you want Claude to produce, and the job is reasonably contained.
For example: “Take this 20-page report and create a 10-slide presentation using this template."
That’s exactly the type of work Sonnet is designed to handle quickly.
Opus 5.5
Opus 5.5 becomes more useful when the instructions leave Claude with more decisions to make. For example: "Review these reports, identify why profitability changed, check whether management's explanation matches the numbers, and prepare a presentation for leadership."
Claude now has to decide what matters, investigate inconsistencies, and form a reasonable analysis before it can create the presentation.
Fable 5.1
Then there's Fable 5.1. Fable was released in September as Anthropic's highest-capability model for coding, research, and long-running problem solving. It’s particularly aimed at situations where Claude may work through a difficult assignment for a long period with relatively little supervision.
Anthropic's examples include complex software work, deep research, scientific work, and multi-step tasks that may run for hours.
Opus 5.5 vs. Fable 5.1: How Close Are They?
The important point is that Opus 5.5 has closed much of the gap. Anthropic says that on most work, Opus 5.5 performs at roughly the level of Fable 5.1 while costing less to run. On its GDPval-AA benchmark, which tests professional work across 44 occupations, Opus 5.5 scored 1,846 compared with 1,735 for Fable 5.1.
Benchmarks are controlled tests, so a higher score doesn’t mean Opus will be better for every prompt. Anthropic itself says the real-world gap between the models can be smaller than benchmark results suggest.
Fable still has a particular role in very difficult, long-running work. Anthropic's testing and early customer reports describe Fable maintaining context across long assignments, independently reprioritizing work, carrying out complex research, and verifying its own progress over extended periods.
Which Claude Model Should You Use?
A practical way to separate them is to think about how much Claude has to figure out for itself. If you know exactly what you want done, Sonnet 5.5 will often be enough. If the problem requires investigation and judgment before Claude can produce an answer, Opus 5.5 is a better fit.
If you are handing Claude a particularly difficult project and expect it to keep working through a large number of steps with limited guidance, Fable 5.1 is where Anthropic is putting its highest-end capabilities.
