Claude Fable 5.2 Completes 41% of Real Freelance Jobs

Anthropic's Claude Fable 5.2 tops the Remote Labor Index at a 41.3% automation rate, double the next best model, according to Scale Labs.

@sharedot

Claude Fable 5.2 Completes 41% of Real Freelance Jobs

By @sharedot · · 6 pages

  • AI Frontier
  • AI Agents
  • Benchmarks
  • Anthropic

Anthropic's Claude Fable 5.2 tops the Remote Labor Index at a 41.3% automation rate, double the next best model, according to Scale Labs.

The record: 41.3% of real freelance work, fully delivered

Anthropic's Claude Fable 5.2 has taken the top spot on the Remote Labor Index (RLI) with a 41.3% automation rate, according to the benchmark's operators, Scale Labs and the Center for AI Safety (CAIS).

That figure means trained human graders judged the model's deliverable at least as good as the human freelancer's — the standard a reasonable client would accept — on 41.3% of the projects. Anthropic's own Opus 5.5 is second at 21.25%, with OpenAI's GPT 6 Astra third at 20.83%.

mlq.ai notes the score was not part of Anthropic's November 9 launch materials and comes instead from the benchmark operators themselves.

Why the leap is startling: 2.5% to 41% in a year

The pace of improvement is the story. mlq.ai reports that when the RLI paper was published in October 2025, the best system, Manus, completed just 2.5% of projects.

Fable 5 led at about 16% in July, and its direct predecessor Fable 5.1 scored 17.92% — so Fable 5.2 more than doubled its own sibling in one version.

According to The Decoder, as reported by mlq.ai, Scale Labs saw the largest gains in audio and music editing, architecture and interior design, and 3D modeling, and nearly 50 of Fable 5.2's successful jobs were ones no agent had ever completed before.

The evidence: 240 real Upwork jobs, human-graded

RLI is not a synthetic test. mlq.ai reports the benchmark's 240 projects came from 358 verified Upwork freelancers and are worth about $144,000 in total — an average of $632.60 and roughly 29 hours of human work each, spanning video editing, CAD, game development and data analysis.

Agents work in a virtual desktop with tools such as Blender, GIMP and Audacity.

Trained experts open each submission in the relevant professional software and compare it against the freelancer's delivered work; CAIS's attempts to automate grading with AI reviewers failed because the reviewers struggled to operate the software and graded too generously, The Decoder reports.

The leaderboard uses 230 private projects, and Scale publishes no confidence interval for the automation rate.

Limits and stakes: industrial design still fails, 3D shines

The results are not uniform. mlq.ai reports that Scale Labs says industrial product design and animation are areas the AI 'can barely handle,' with typical failures including amateur-quality work, incomplete or broken files, and contradictory designs.

The scores also measure completion on a fixed sample rather than estimating what share of the remote labor market AI can automate, and Anthropic has not commented on the result.

Meanwhile, Crypto Briefing reports the same model now leads Artificial Analysis' AA-3D benchmark at 1271 Elo — 67 points clear of Claude Sonnet 5.5 and 103 ahead of GPT-6 Astra — while costing 20% less than its predecessor.

For developers, a single model that tops both real-work and 3D benchmarks at a lower price resets the agent tooling calculus.

Sources

  1. mlq.ai › Anthropic's Claude Fable 5.2 completes 41% of real freelance jobs on the Remote Labor Index, double the next model
  2. cryptobriefing.com › Claude Fable 5.2 takes first place on Artificial Analysis' AA-3D benchmark

More on AI Frontier