FirmCORe - Role Direction: leaderboard

Predict which firm plays which role in a detected FirmCORe collaboration opportunity. Macro-F1, higher is better. Source: arxiv.org/abs/2609.17128.

Metric: Macro-F1 (%). Source: arxiv.org. Saturation forecast: Around October 2027. 17 models tracked.

Top models

#ModelScoreOverall rank
1Qwen 3.7 Plus67.04#98
2Gemma 4 26B66.52#322
3GLM-5.163.72#109
4Qwen 3.6 35B A3B63.38#238
5Gemma 3 12B61.02#674
6GLM-5.259.87#80
7Qwen3 Coder59.54#304
8Qwen 3.6 Max58.61#97
9Qwen 3.6 27B56.11#182
10Qwen 3 4B54.89#834
11Llama 3.1 8B52.25#1130
12Qwen 3 8B52.1#678
13Llama 3.2 3B22.25#1376

No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.

Interactive version: theaggregate.ai/benchmark?slug=firmcore-role-direction · How It Works · Data refreshed daily, snapshot 2026-11-05.