AgentS4D - Task Completion (Codex CLI): leaderboard
Task completion on the 328 risk-injected AgentS4D (Zhejiang University of Technology, 2026) workspace cases run in the Codex CLI harness. The score is the percent of runs that satisfy the case-specific completion check for the authorized task despite the injected attacker content; higher is better.
Metric: Task completion rate (%): runs whose case-specific completion predicate for the authorized task is satisfied, over the 328 risk-injected cases run once in the Codex CLI (0.142.5) harness; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4 Pro | 96.65 | #188 |
| 2 | GPT-5.5 | 94.82 | #49 |
| 3 | Gemini 3.1 Pro (Preview) | 91.46 | #94 |
| 4 | Qwen 3.7 Plus | 91.46 | #207 |
| 5 | MiniMax-M3 | 89.02 | #269 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=agents4d-task-completion-codex-cli · How It Works · Data refreshed daily, snapshot 2026-11-13.