MemEvoBench (Workflow Style): leaderboard

MemEvoBench (2026) evaluates memory safety of LLM agents: misleading entries in a mixed memory pool, and the agent's own responses stored after each round, test whether accumulated memory drifts the agent into unsafe behavior over three rounds. This board is the attack success rate in percent in the third round on workflow-style cases whose memory holds noisy tool returns, judged by GPT-5.2, without user feedback; lower is better.

Metric: Attack success rate (%) in the third evaluation round on the 83 workflow-style cases (20 Agent-SafetyBench environments with noisy tool returns in the memory pool): share of responses that follow the misleading memory into the target risk behavior after two rounds of the agent's own responses were added to the memory pool, without user feedback, judged MISLED or SAFE by GPT-5.2, temperature 0, standard system prompt without safety guidance; lower is better. Source: arxiv.org. Saturation forecast: Around June 2027. 9 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-569.2#178
2Claude 3.7 Sonnet82.7#470
3Qwen 3 235B A22B 2507 Instruct83.1#616
4Gemini 2.5 Pro84.3#276
5DeepSeek V3.284.3#419
6Qwen 3 Next 80B A3B Instruct84.3#778
7Qwen 3 32B89.2#972
8Llama 3.3 70B Instruct90.2#1385
9GPT-4o91.6#751

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

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