MultiSessionCollab - MMLU: leaderboard

Metric: Task success (%; accuracy of the simulated user's final draft answer, MMLU problems, 100 simulated users x 20 sessions, agent without memory). Source: arxiv.org. Saturation forecast: Around October 2027. 4 models tracked.

Top models

#ModelScore
1Llama 3.3 70B Instruct51.26
2Llama 3.1 8B Instruct50.5
3GPT-OSS-20B49.62
4Qwen 2.5 7B Instruct47.85

Interactive version: theaggregate.ai/benchmark?slug=multisessioncollab-mmlu · How It Works · Data refreshed daily, snapshot 2026-09-26.