Artificial Analysis Refreshes Coding Agent Index With FrontierCode v1.1 and Terminal-Bench 4.0

Artificial Analysis released version 2.1 of its Coding Agent Index on November 14, 2026, with Anthropic's Claude Fable 5.2 and Claude Opus 5.5 tied for first place at a composite score of 58. Crypto Briefing reported the release.

How the Coding Agent Index v2.1 Works

The index evaluates combinations of a model and an agent harness, the software scaffolding that lets an AI read files, run commands and take actions over multiple steps. Version 2.1 spans approximately 290 tasks across three benchmarks, each carrying equal weight in the final composite score.

FrontierCode covers software engineering work on real open-source repositories. The index now uses FrontierCode v1.1, built by Cognition, which includes 100 tasks and grades work on whether repository maintainers would merge it. Terminal-Bench measures agentic terminal use, and the index now runs Terminal-Bench 4.0 from the Laude Institute, which contains 66 tasks. SWE-Atlas-QnA-AA tests repository question and answering, whether an agent can understand an unfamiliar codebase well enough to answer questions about it accurately.

Version 2.0 of the index launched on October 23, 2026 and introduced the three-part structure built around FrontierCode, Terminal-Bench and SWE-Atlas-QnA-AA. Version 2.1 keeps that framework and refreshes the underlying benchmarks. Claude Fable 5.2 is the newer of the two co-leaders; Anthropic launched it on November 9, 2026, positioning it for complex coding and agentic challenges. Claude Opus 5.5 carried the top score of 61.7 in v2.0 and now sits at 58 alongside its newer sibling.

Artificial Analysis's Benchmark Record

Artificial Analysis is an independent AI benchmarking firm co-founded by George Cameron and Micah Hill-Smith, according to akta.pro company records. The same records show the company tracks 542 models across various benchmarks, with 27 of these currently ranked in the Intelligence Index v4.1. The records list one funding round, a seed round in January 2025 with one investor.

Earlier Artificial Analysis Launches

Artificial Analysis launched the HLE-Diamond leaderboard on October 27, 2026, with Google's Gemini 4 Carbon leading at 69.1% overall accuracy, Crypto Briefing reported. On November 10, 2026, Google DeepMind and Artificial Analysis released the Multimodal Knowledge Agent benchmark, testing AI agents on visual tasks such as blueprints and clinical scans, Crypto Briefing reported. Artificial Analysis launched its Cyber Index v1 on September 25, evaluating AI models on enterprise vulnerability defense across three benchmarks, Crypto Briefing reported.

Cognition's Devin Apps Release

Cognition launched Devin Apps on November 13, 2026, a feature enabling users to create live, private web applications hosted on Devin's cloud computer, Investing.com reported. The apps run on Devin's machine with access to MCPs and codebases, are billed by uptime, and are never public.