NEWS · MODELS · #3343
Anthropic's Fable 5.2 completes 41.3% of freelance jobs on Remote Labor Index, outpacing rivals
Anthropic's Fable 5.2 achieved a 41.3% completion rate on the Remote Labor Index (Scale Labs and CAIS), delivering professional‑quality results for many real Upwork projects and substantially outperforming prior models and competitors (Opus 5.5 at 21.25%, GPT 6 Astra at 20.83%). The RLI evaluates 240 real assignments in virtual professional workspaces and uses human expert reviewers to judge whether AI outputs meet paying‑client standards.
KEY POINTS
- Anthropic's Fable 5.2 achieved a 41.3% completion rate on the Remote Labor Index (Scale Labs and CAIS), delivering professional‑quality results for many real Upwork projects and substantially outperforming prior models and competitors (Opus 5.5 at 21.25%, GPT 6 Astra at 20.83%).
- The RLI evaluates 240 real assignments in virtual professional workspaces and uses human expert reviewers to judge whether AI outputs meet paying‑client standards.
- This demonstrates rapid, practical gains in AI agents' ability to complete end‑to‑end professional work, affecting labor, productization, and how capabilities are evaluated.
WHY IT MATTERS
This demonstrates rapid, practical gains in AI agents' ability to complete end‑to‑end professional work, affecting labor, productization, and how capabilities are evaluated.