Agents' Last Exam v1.1 - OpenClaw CLI Partial Credit: leaderboard
ALE v1.1 evaluates agents on 122 long-horizon professional tasks in the OpenClaw CLI harness. Partial Credit is the mean graded score (%) across all 122 tasks, including failed or missing runs as zero; timed-out runs retain their graded scores. Higher is better. Exact model and effort settings stay separate from other harnesses and legacy ALE v1.
Metric: Partial Credit (%). Source: agents-last-exam.org. Saturation forecast: Rough model projection: around 2026. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | deepseek-deepseek-v4-1-flash-1017 (ALE Run: harness=OpenClaw%20CLI; effort=max; source-model=deepseek-deepseek-v4-1-flash-1017) | 66.64 | |
| 2 | deepseek-deepseek-v4-1-flash-1017 (ALE Run: harness=OpenClaw%20CLI; effort=high; source-model=deepseek-deepseek-v4-1-flash-1017) | 65.4 |
Interactive version: theaggregate.ai/benchmark?slug=agents-last-exam-v1-1-openclaw-cli-partial-credit · How It Works · Data refreshed daily, snapshot 2026-11-05.