Claude Fable 5 is amazing at biology, but won't answer basic biology questions
·2 min read·Beginner
“
Anthropic just dropped Claude Fable 5, hyping it as their most powerful model ever released, with exceptional skills in biology. Here's the kicker: ask it a basic biology question—the kind any high school student should handle—and it punts the whole thing to its predecessor instead.
In 30 seconds
01Anthropic promotes Claude Fable 5 as excellent in biology, but it refuses basic high school questions.
→
💡
What this means for you
If you're thinking of using Fable 5 for biology questions—even basic ones—there's a real chance it'll fail you. The bigger message is more unsettling: the promises companies make about what their models can do might not match what actually happens when you use them in the real world.
Thought slapping 'AI' next to a company name guaranteed its stock would soar? Well, the market had a bitter surprise this year.
·1 min·2·Beginner
02
The model redirects to Claude Opus instead of answering elementary biology.
03Gap between marketing claims and real performance raises doubts about official benchmarks.
Yes, you read that right. Anthropic rolled out Fable 5 with huge fanfare as their most powerful model yet, specifically touting its excellence in biology and other sciences. But when people actually tried asking it basic biology questions—the kind of thing like "how does photosynthesis work?" or "what's the difference between mitochondria and chloroplasts?"—the model refused to answer. Not because it can't: it literally told people "hey, this is a job for Claude Opus" (the previous model) and bailed. It's like buying a sports car and finding out it won't drive on the highway.
The whole thing gets weirder because we're not talking about Nobel Prize–level questions here. We're talking high school biology. Homework assignment stuff. According to tests that have been run, Fable 5 systematically refuses to answer these questions, preferring to kick everything back to Opus. Whether this is an intentional design choice (some kind of safety guardrail?), a bug, or just a confused way of handling requests outside its "comfort zone" is genuinely unclear.
📬 Enjoying this article?
Get the best AI news every week, straight to your inbox.
And here's where the irony hits: Anthropic spent all this effort emphasizing how Fable 5 improved in science and biology specifically. They literally said in their press release that the model excels in these areas. Yet according to real-world testing, when you put the model in front of actual biology scenarios, it chokes. It's like a restaurant advertising their hot new Italian chef and then when you order pasta they're like "sorry, our new chef doesn't do pasta, we'll get the old guy."
The issue isn't just the awkwardness of a model that promises something and doesn't deliver. It raises a pretty uncomfortable question about the nature of the benchmarks these companies use to test their models. If the model tanks on basic biology questions despite promising to be good at biology, how can we really trust what companies tell us their models can do? It's a bit like reading five-star reviews on a product and then discovering it doesn't work for 50% of actual use cases.
For now, the mystery sits there. We don't know if Anthropic is trying to keep processing costs down by offloading harder questions to Opus, if there's a routing bug, or if the model just has constraints that weren't clearly communicated to the public. What we do know is that the gap between what's promised and what actually works is bigger than anyone wants to admit.
While the tech world was buzzing about OpenAI, Anthropic made its move. They just dropped Opus 5, a model they claim is almost as good as their legendary Fable 5.
NUOVO
