Anthropic's Fable 5.2 tops freelance AI benchmark, still fails 58%
Fable 5.2 leads Scale's Remote Labor Index, yet more than half of its deliverables don't measure up to professional work.
Advertising disclosure: we may earn a commission when you join an operator via links on this page, at no cost to you. This never dictates our ratings. How we make money
Key takeaways
- Fable 5.2 ranks first on Scale's Remote Labor Index.
- The index grades complete deliverables against professional reference work.
- The headline figure: the model fails 58% of the tasks.
- Even the top model still falls short of human freelancers on most jobs.
Anthropic's Fable 5.2 now holds the top spot on Scale's Remote Labor Index, a benchmark for AI on freelance-style work. It still fails 58% of the tasks, per CryptoSlate's report.
What the index measures
This isn't a trivia quiz or a puzzle set. The Remote Labor Index takes a model's complete deliverable and compares it with reference work produced by professionals. That's a harder test, and a cleaner one. A finished product either holds up or it doesn't.
Leading isn't passing
Fable 5.2 ranks first. Yet a 58% failure rate means that on most tasks, its output wasn't good enough to sit beside the professional version. Best of a weak field? Not the same as ready to replace a freelancer.
The source material I have is thin, so I can't tell you how the other models scored or which task types tripped Fable up.
Why it matters
Crypto is crowded with projects pitching AI agents that supposedly do real work for you, and plenty of tokens ride on that story. This benchmark is a handy reality check. If the leading model misses on more than half of complete jobs, agents built on weaker models will likely do worse. So be sceptical of bold claims about autonomous AI labor, especially when a token's attached. Ask for results on full deliverables, not demos.
The better question is how fast that 58% shrinks.
Source: CryptoSlate
Top Crypto Casinos Right Now
🇬🇧 Showing sites that accept players from United Kingdom Change country
Advertising disclosure: we may earn a commission when you join an operator via links on this page, at no cost to you. This never dictates our ratings. How we make money
More info Less info
Why we chose it: A US-oriented casino that turns away only New Jersey. The cashier takes crypto, but the site publishes fewer specifics than any operator we list.
- KYC
- Required
Restricted countries: New Jersey (US).
More info Less info
Why we chose it: The LatAm door of the BetNow platform: Spanish-first, a 200% welcome at 20x rollover, and a weekly 2% rebate. Crypto rails unconfirmed — check before depositing.
- KYC
- Required
Restricted countries: Australia, Costa Rica, Central African Republic, Congo - Brazzaville, Eritrea, France, Guinea-Bissau, Iraq and 15 more.
Mia Chen
Mia digs into DeFi and on-chain data, translating protocol mechanics into plain English for everyday readers.
Related News
DeepMind's AI agents ran a four-day crypto test with ~20 human messages
Near AI breach exposes user details and data from 18 AI agents
CrowdStrike Builds a Kill Switch for AI Agents, Plans to Buy Fiddler AI
