PI-AutomationBench direct mode (strict)
Agents · 2026-11-03
The strict score Prime Intellect's chart reports beside PI-AutomationBench's partial score in 'mode: direct', with Prime Agent. The post doesn't define strict further or state the task count.
Benchmark source: Primeintellect
Top models (higher is better)
| Model | Score | Source |
|---|---|---|
| DeepSeek V4.1 Flash 1017 | 87.4 | Prime Intellect, Learnings from Auditing Public Evaluations · Prime Intellect |
| Claude Opus 5.5 | 73.8 | Prime Intellect, Learnings from Auditing Public Evaluations · Prime Intellect |