---
format: "aidr-story-markdown/v1"
id: "a1c2dc69bcd6a767c0248d3ce5e4cfb57bd636c3c66c8605732cf3ea4f01f52e"
canonical_url: "https://aidr.today/a1c2dc69?lang=en"
title: "Claude Fable 5.2 Leads RoboWild Robot Benchmark with 13.3% Accuracy"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-21T03:43:43.000Z"
category: "Models"
topics: ["anthropic","llm","agent"]
source_urls: ["https://marketbrief.now/ai/claude-fable-52-leads-robowild-robot-benchmark-with-133percent-accuracy-1f8b3ec0","https://huggingnews.com/ai/claude-fable-52-leads-robowild-robot-benchmark-with-133percent-accuracy-1f8b3ec0"]
summary: "Claude Fable 5.2 led the RoboWild robot benchmark with 13.3% accuracy, followed by GPT-6 Astra at 11.6%. The XLangNLP test evaluates how AI models generalize to real-world robot use across 341 tasks and 4,480 objects."
---

# Claude Fable 5\.2 Leads RoboWild Robot Benchmark with 13\.3% Accuracy

> [Open the canonical story](<https://aidr.today/a1c2dc69?lang=en>)

**Published:** 2026-11-21T03:43:43.000Z
**Category:** Models
**Topics:** anthropic, llm, agent

## Summary

Claude Fable 5\.2 led the RoboWild robot benchmark with 13\.3% accuracy, followed by GPT\-6 Astra at 11\.6%\. The XLangNLP test evaluates how AI models generalize to real\-world robot use across 341 tasks and 4,480 objects\.

## Sources

- [Story source](<https://marketbrief.now/ai/claude-fable-52-leads-robowild-robot-benchmark-with-133percent-accuracy-1f8b3ec0>)
- [Story source](<https://huggingnews.com/ai/claude-fable-52-leads-robowild-robot-benchmark-with-133percent-accuracy-1f8b3ec0>)

