EMem-Bench - Dynamic Tracking: leaderboard

Metric: Success rate (%; the 1052 Dynamic Tracking episodes, acting on the latest location or state of an object that moved or changed; full-context condition: the complete interaction history as sanitized text plus the current 500 x 500 RGB view, one primitive action per call with deterministic decoding and at most eight target-task steps in AI2-THOR and ProcTHOR homes; an episode succeeds only if the simulator reaches the target terminal state). Source: arxiv.org. Saturation forecast: Around October 2027. 17 models tracked.

Top models

#ModelScore
1Gemini 3 Flash53.9
2Gemini 2.5 Pro51.6
3Qwen 3.6 27B49.5
4GPT-5.446.7
5Qwen 3 VL 4B Instruct37.3
6InternVL3-8B31.8
7Qwen 3 VL 8B Instruct26.6
8GPT-5.4 Mini25.8
9Qwen 3 VL 32B Instruct20.3
10InternVL3-38B17

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