# Artificial Analysis Coding Agent Index v2.0 ranks Claude Opus 5.5 first at 61.7

> Original title: Coding Agent 榜单升级到 v2.0：FrontierCode 入场，Claude Opus 5.5 以 61.7 领跑

- Source: meng shao
- Published: 2026-10-23T14:33:44.000Z
- Bad Influence: https://badinfluence.news/items/cst3dtn4c1k0b577alfkqphx3
- Original: https://x.com/shao__meng/status/2113640018292207934

## Summary

Artificial Analysis released v2.0 of its Coding Agent Index, which evaluates models with their agent harnesses across 290 tasks. Claude Code with Claude Opus 5.5 (max) leads at 61.7, and FrontierCode v1.1 replaces DeepSWE. The mean score for the same 21 configurations fell from 54.8 to 47.4 under the new version.

## Post

Coding Agent 榜单升级到 v2.0：FrontierCode 入场，Claude Opus 5.5 以 61.7 领跑

@ArtificialAnlys 发布 Coding Agent Index v2.0：用 @cognition 的 FrontierCode v1.1 替换 DeepSWE，并收紧仓库问答评分。榜单评的是「模型 + harness」组合，不是裸模型；当前第一是 Claude Code + Claude Opus 5.5（max）61.7。
https://artificialanalysis.ai/agents/coding-agents

榜单评什么
三套基准等权平均，共 290 题，每题跑 3 次再取均值：
· FrontierCode v1.1（Cognition）：100 题，测试在真实开源仓库改代码，按「维护者会不会合」打分
· Terminal-Bench 4.0：66 题，测试终端里的多步 agent 任务
· SWE-Atlas-QnA-AA（Scale AI 题 + AA 改判）：124 题，测试读仓库、答技术问题

相对 v1.5 的两处改动
1. FrontierCode 换下 DeepSWE。 题集不公开。失败任一 blocking 检查 → 0 分；全过 → 70%–100%（看 non-blocking）。轨迹若被扫到去拉上游解（PR、merge commit、镜像等）→ 也记 0。
2. SWE-Atlas-QnA → SWE-Atlas-QnA-AA。 124 题不变；重写了 1,306 条 rubric 里的 127 条；裁判改为 GPT-6 Luna（xhigh），每条准则投 5 票；数据集与裁判配置已发到 Hugging Face。

# 关键结果
综合指数前四：
· Claude Code + Opus 5.5（max）61.7
· Claude Code + Sonnet 5.5（max）59.3
· Muse Code + Muse Wonder Full（max）58.9
· Claude Code + Fable 5.1（max + fallback）57.6
前四相差仅约 4.1 分，其中三席是 Claude Code。首个 OpenAI 组合是 Codex + GPT-6 Astra，54.7。

FrontierCode 单项（Opus 拉开差距处）：
· Opus 5.5：56.0%
· Codex + DeepSeek V4.1 Flash：52.6%
· Codex + GPT-6.1 Sol：49.4%
· Codex + GPT-6 Astra：49.3%
其余 24 个配置里有 15 个挤在 43%–49%

仓库问答（SWE-Atlas-QnA-AA）前几名：
· Muse Wonder Full：68.8%
· Claude Fable 5.1：66.4%
· Opus 5.5：65.9%
· Sonnet 5.5：64.2%

同一批 21 个配置在新旧 Index 上对比：均值从 54.8 → 47.4，20/21 下降；主因是 FrontierCode 均值约 45.1%，远低于 DeepSWE 的 65.6%；问答侧均值再掉约 1.7 分（58.8% → 57.1%）。Terminal-Bench 4.0 这轮没换。

成本前沿（按 token API 计，非订阅价）
· Codex + DeepSeek V4.1 Flash：47.3，$0.39
· Codex + GPT-6.1 Sol：54.1，$1.50
· Codex + GPT-6 Astra：54.7，$7.04
· Claude Code + Opus 5.5：61.7，$11.42
第一名大约是最便宜前沿点的 29 倍 单价，多约 14.4 分。

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHVUoXgAaEAAdN4x.jpg&mode=thumb&exp=1793404800&sig=a06b42d3653f5afe)

## Quoted post @ArtificialAnlys

> Announcing Artificial Analysis Coding Agent Index v2.0: FrontierCode v1.1 from Cognition replaces DeepSWE, SWE-Atlas-QnA grading is revised, and Claude Opus 5.5 in Claude Code leads at 61.7
> 
> Today we are launching v2.0 of the Coding Agent Index. The Index evaluates models together with their agent harnesses, now across 290 tasks covering software engineering, terminal use, and repository Q&A, with the three benchmarks weighted equally.
> 
> Changelog (Index v1.5 → Index v2.0):
> 
> → FrontierCode v1.1 replaces DeepSWE v1.1. FrontierCode is a benchmark from @cognition with a private test set: 100 software engineering tasks on real open-source repositories, graded on whether a maintainer would merge the change. An attempt that fails a blocking check scores zero, and one that passes them all scores between 70% and 100% depending on the non-blocking checks it meets. We also scan each trajectory with a modified version of Cognition's rule-based scanner, and any attempt flagged for accessing solution material scores zero.
> 
> → Revised SWE-Atlas-QnA grading, now SWE-Atlas-QnA-AA. Scale AI's 124 repository Q&A tasks are unchanged. We reworded 127 of the 1,306 rubric criteria for more consistent verdicts and moved to GPT-6 Luna (xhigh) as the judge, with five votes per criterion. The dataset and judge configuration are published on Hugging Face.
> 
> Key results:
> 
> ➤ Claude Opus 5.5 leads the Index: Claude Code with Claude Opus 5.5 (max) scores 61.7, followed by Claude Code with Claude Sonnet 5.5 (max) at 59.3, Muse Code with Muse Wonder Full (max) at 58.9 and Claude Code with Claude Fable 5.1 (max with fallback) at 57.6.
> 
> ➤ FrontierCode v1.1 is where Opus 5.5 pulls ahead: Opus 5.5 scores 56.0% on FrontierCode, ahead of Codex with DeepSeek V4.1 Flash (max) at 52.6%, Codex with GPT-6.1 Sol (max) at 49.4% and Codex with GPT-6 Astra (max) at 49.3%. Below the top, FrontierCode scores are tightly packed: 15 of the 24 configurations on the Index score between 43% and 49%.
> 
> ➤ SWE-Atlas-QnA-AA grading is stricter and more consistent: across the 21 configurations evaluated on both Index versions, the mean score falls 1.7 points, from 58.8% to 57.1%, with 14 of 21 moving down. Individual configurations move by up to 7 points: Muse Code with Muse Spark Open (xhigh) falls 7.0, Kimi K3 in Kimi Code CLI and Codex with GPT-6 Luna (max) 6.2, and Muse Code with Muse Wonder (max) 5.6, while Codex with GPT-6.1 Sol (max) gains 2.1.
> 
> ➤ Four configurations sit on the cost frontier: Codex with DeepSeek V4.1 Flash (max) scores 47.3 at $0.39 per task, Codex with GPT-6.1 Sol (max) 54.1 at $1.50, Codex with GPT-6 Astra (max) 54.7 at $7.04 and Claude Code with Claude Opus 5.5 (max) 61.7 at $11.42. Costs are pay-per-token API costs, not subscription prices.
> 
> We will keep updating the Coding Agent Index incrementally, with more evaluations and agents to come.

https://x.com/ArtificialAnlys/status/2113424551396934134