MMOOC - Partial Answerability (Multiple Choice): leaderboard

Metric: Shifted in-context score (%; mean of accuracy on the answerable Partial Answerability Multiple Choice questions and answer rationality, the mean 0-1 judge score of the answer's correctness and evidence consistency in steps of 0.25; judge scores averaged over GPT-5.6, Claude Opus 5 and DeepSeek-V4-Pro). Source: arxiv.org. Saturation forecast: Around December 2026. 18 models tracked.

Top models

#ModelScore
1Qwen 3.5 122B A10B90.5
2Llama 4 Maverick86.5
3Qwen 3.5 27B85.5
4Gemma 4 26B85.25
5Gemma 4 31B84.75
6Ministral 3 14B78.5
7GPT-4o76.25
8O375.25
9O173.25
10Ministral 3 8B71.75
11InternVL3-8B71.5
12Gemini 3.1 Pro (Preview)58.75
13Claude Opus 4.645.25

Interactive version: theaggregate.ai/benchmark?slug=mmooc-partial-answerability-multiple-choice · How It Works · Data refreshed daily, snapshot 2026-09-29.