MARCA (English, Orchestrator): leaderboard

MARCA (2026) tests LLM web search on 52 manually authored multi-entity questions asked in English, in the Orchestrator framework (the model delegates sub-questions to subagents that call web_search and web_scrape). A GPT-4.1 judge checks each answer against a manually validated checklist; checklist accuracy (%) averaged over three runs, higher is better.

Metric: Checklist accuracy (0-1, times 100): share of a question's checklist items (expected entities and attributes) that a GPT-4.1 judge marks satisfied, averaged over MARCA's 52 English questions and three runs, Orchestrator framework (the model delegates sub-questions to subagents that call web_search and web_scrape); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 14 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Pro89.7#145
2GPT-5.288.1#185
3Sabia-483.2#440
4Gemini 3 Flash83.1#175
5Kimi K283#443
6GPT-5 Mini82.7#364
7GPT-4.177.9#505
8Qwen 3 235B A22B75.3#628
9Sabiazinho-472.1#668
10GPT-4.1 Mini71.9#789
11Gemini 2.5 Pro70.1#276
12Gemini 2.5 Flash54#463
13Qwen 3 30B A3B52#1156
14Gemini 2.5 Flash Lite31.7#964

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

Interactive version: theaggregate.ai/benchmark?slug=marca-english-orchestrator · How It Works · Data refreshed daily, snapshot 2026-11-21.