# The Last Ones Average Progress (The Last Ones steps)

> Average step reached on a 32-step long-horizon cyber range.

Canonical page: https://benchlm.ai/benchmarks/lastonescyberrangesteps

- Category: [Agentic](/agentic)
- Last updated: September 29, 2026

## About The Last Ones steps

- Year: 2026
- Tasks: 32-step long-horizon cyber range
- Format: Average step reached
- Difficulty: Long-horizon cyber operations
- Paper: [UK AISI and CAISI Preliminary Assessment of Kimi K3's Cyber Capabilities](https://www.nist.gov/news-events/news/2026/07/uk-aisi-caisi-preliminary-assessment-kimi-k3s-cyber-capabilities)

The joint assessment reports average progress through a 32-step cyber range. BenchLM stores the published step number as display-only external evidence.

The Last Ones steps is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (1 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Kimi K3](/models/kimi-k3) | Moonshot AI | 17% |

## FAQ

### What does The Last Ones steps measure?

Average step reached on a 32-step long-horizon cyber range.

### Which model scores highest on The Last Ones steps?

Kimi K3 by Moonshot AI currently leads with a score of 17% on The Last Ones steps.

### How many models are evaluated on The Last Ones steps?

1 AI models have been evaluated on The Last Ones steps on BenchLM.

### Does The Last Ones steps affect BenchLM's overall score?

Not directly. The Last Ones steps is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.