Anthropic has secured the top position on the Epoch Capabilities Index with its newly released Claude Fable 5 model, overtaking OpenAI’s GPT 5.5 Pro and ending a period of more than a year during which Anthropic did not lead the benchmark. According to the latest rankings published by Epoch AI, Claude Fable 5 is currently ranked first among 198 evaluated models with an Epoch Capabilities Index score of 161, while GPT 5.5 Pro follows closely in second place with a score of 160. The one point difference may appear narrow, but it marks a notable shift in the competitive landscape of advanced artificial intelligence models. Epoch AI stated that Claude Fable 5 achieved a new high score on the index, making it the highest ranked model currently assessed under its capability evaluation framework. The achievement represents Anthropic’s first return to the top of the index since 2024, when Claude 3.5 Sonnet was among the company’s leading models.
The Epoch Capabilities Index is designed to provide a broad measurement of AI model performance by combining results from a wide range of benchmark evaluations into a single capability score. Rather than focusing on one specific skill area, the index aims to assess general model competence across multiple domains and maintain comparability as individual benchmarks become saturated over time. This methodology allows researchers and industry observers to track long term progress among competing AI systems and evaluate overall capability trends. By aggregating results from various benchmark categories, the index seeks to present a more comprehensive view of model performance than standalone tests. Claude Fable 5’s leading position therefore reflects strong results across several areas rather than dominance in a single benchmark, highlighting the model’s broad capabilities across different tasks and evaluation environments.
Anthropic introduced Claude Fable 5 on June 9, 2026, describing it as a Mythos class model that has been adapted for general availability with additional safety measures. The company stated that the model is its most capable publicly available AI system and performs particularly well in software engineering, knowledge work, scientific research, vision related tasks, and extended complex reasoning activities. Anthropic also noted that certain sensitive requests may automatically be routed to Claude Opus 4.8 due to additional safeguards incorporated into the platform. Beyond the overall benchmark ranking, Epoch AI reported strong mathematical performance from Claude Fable 5, particularly on FrontierMath evaluations. According to the organization, the model achieved scores of 87 percent across FrontierMath Tiers 1 through 3 and 88 percent on Tier 4 assessments. These results were described as further evidence of Anthropic’s continued progress in advanced mathematical reasoning and problem solving capabilities.
The benchmark milestone arrives during a challenging period surrounding access to Anthropic’s latest models. Following the launch of Claude Fable 5 and Claude Mythos 5, Anthropic disclosed that access to both systems had become unavailable after a directive issued by the U.S. government. The company said it had suspended access to the models while working to restore availability as quickly as possible. Despite those access limitations, Claude Fable 5’s performance on the Epoch Capabilities Index has positioned Anthropic ahead of OpenAI on one of the industry’s closely watched capability rankings. While the margin separating Claude Fable 5 and GPT 5.5 Pro remains extremely small, the latest results underscore the increasingly competitive nature of the artificial intelligence sector, where leading developers continue to push for incremental gains across a growing range of benchmarks and real world applications.
Follow the SPIN IDG WhatsApp Channel for updates across the Smart Pakistan Insights Network covering all of Pakistan’s technology ecosystem.
