ACE-Bench (Azure SDK, MCP Documentation): leaderboard

Azure SDK coding on ACE-Bench (2026), the Azure SDK Coding Evaluation Benchmark: tasks built from official documentation examples in Java, JavaScript or TypeScript, C# and Python, checked without execution by regex rules and a reference-grounded LLM judge. Strict pass rate in percent when the agent may query the Microsoft Learn documentation server; higher is better.

Metric: Strict pass rate (%) on ACE-Bench (353 Azure SDK coding tasks in Java, JavaScript/TypeScript, C# and Python built from official documentation examples; a task passes only if every atomic criterion holds: regex checks of required API usage and reference-grounded LLM-judge checks); same lightweight mcp-use coding agent for every model, augmented: the agent may query the Microsoft Learn MCP documentation server before answering; higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Grok 468.7#319
2Claude Opus 4.565.3#160
3Claude Sonnet 4.563.7#261
4GPT-5.163.2#240
5Grok Code Fast 158.5#703
6Claude Haiku 4.558#583
7Claude Opus 4.153.6#278
8GPT-553.5#178
9GPT-4.151.1#505
10Grok 4 FastNon-reasoning50.9#447
11GPT-5 Mini49.6#364

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=ace-bench-azure-sdk-mcp-documentation · How It Works · Data refreshed daily, snapshot 2026-11-21.