Artificial Analysis releases Coding Agent Index v2.1: Claude Fable 5.2 and Opus 5.5 share the lead at 58

Artificial Analysis released v2.1 of its Coding Agent Index, which scores models together with their agent harnesses. It upgrades FrontierCode from 1.1 to 2.0 and Terminal-Bench from 4.0 to 4.1. The scores are Artificial Analysis's own.

What changed:

• FrontierCode-AA 2.0 replaces FrontierCode v1.1. It is Artificial Analysis's implementation of Cognition's FrontierCode 2.0: 132 real-world coding tasks, graded on whether a repository maintainer would merge the agent's change.
• Terminal-Bench 4.1 replaces 4.0: the same 66 tasks, with updated verifiers that Artificial Analysis says improve reliability and resistance to reward hacking.
• SWE-Atlas-QnA-AA stays in the Index.
• Scores are lower in v2.1, by about 5 points on average, driven by FrontierCode-AA 2.0. Codex with GPT-6.1 Terra (max) is the only configuration that did not fall, staying within 1 point of its earlier score.

Results:

• Claude Fable 5.2 and Claude Opus 5.5 share the lead at 58. Fable 5.2 scores 58 at both xhigh and high in Claude Code; Opus 5.5 (max) scores 58 in Claude Code and, paired with GPT-6 Luna High, in Cognition's multi-model harness Devin Fusion CLI.
• Fable 5.2 (max) scores 56, as more of its attempts hit task time limits. Claude Sonnet 5.5 (max, 56) in Claude Code and Gemini 4 Carbon (high, 55) in Antigravity CLI follow.
• Cost per task varies about 30x among configurations scoring 48 or more: from $1.51 for Codex with GPT-6.1 Sol (max, 52) to $44.22 for Claude Code with Fable 5.2 (max, 56).