# Moonshot removed cyber data from Kimi K3.1 Pro before its open release

> Original title: Moonshot deliberately weakened Kimi K3.1 Pro’s hacking ability before the open release.

- Source: Rohan Paul
- Published: 2026-10-23T19:26:21.000Z
- Bad Influence: https://badinfluence.news/items/pz6we24jd7lrkwzvkw9m51j29
- Original: https://x.com/rohanpaul_ai/status/2113713658887102788

## Summary

Moonshot AI says it removed cyber-related data from mid-training and reinforcement learning for Kimi K3.1 Pro before releasing its weights. On its internal xdev-bench-hard, pass@3 on exploit tasks fell from 90.5% in the uncut version to 81.0%, still above K3's 33.3%. Full weights are due by November 4 for independent reruns.

## Why it matters

The source explains why Moonshot removed cyber-related training data and how that changed exploit performance, which is useful context for judging open-weight release safety.

## Post

Moonshot deliberately weakened Kimi K3.1 Pro’s hacking ability before the open release.

the uncut version solved 90.5% of in-house exploit tasks, while the released model solved 81%, still well above K3’s 33.3% and behind DeepSeek V4.1-Flash at 85.7% by Moonshot’s count.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHVVrPXxa8AApat9.jpg&mode=thumb&exp=1793404800&sig=a62517f20da5f950)

## Quoted post @rohanpaul_ai

> Moonshot AI’s first update, the new Kimi K3.1 Pro
> 
> > Kimi K3.1 Pro scores 65.7% on Terminal-Bench 4.0, 0.7 points behind Claude Opus 5.5 and 7.5 ahead of GPT-6 Astra, at a lower average cost per task than either rival’s top runs.
> 
> > Moonshot saw exploit skill climb as post-training scaled, so it removed cyber-related data from mid-training and RL; on its internal xdev-bench-hard, pass@3 fell from 90.5% to 81.0%.
> 
> > The filtered model still solves 81.0% of those exploitation tasks against K3’s 33.3%, which Moonshot reads as general coding gains carrying into cyber work, while a third-party run of GPT-5.6 Sol solved 100%.
> 
> > On Moonshot’s internal hallucination stress test with tools, false claims fell from 23.4% for K3 to 8.3%, still above GPT-6 Astra’s 5.5% but well below Claude Opus 5.5’s 15.0%.
> 
> > The biggest jump is on Moonshot’s simulated business-operation test, where scores rose from 8.9 to 37, and K3.1 Pro also tops OfficeQA Pro v2 at 64.4 against Opus 5.5’s 62.2.
> 
> > Gains are uneven: K3.1 Pro scores 38.1 on Terminal-Bench Science against GPT-6 Astra’s 68.1 and Opus 5.5’s 63.3, while leading Agents’ Last Exam at 43.2 and Humanity’s Last Exam with tools at 70.0.
> 
> > K3.1 Pro and a Highspeed variant running 2-3x faster are live in Kimi’s apps and API, with full weights due by November 4 for independent reruns.

https://x.com/rohanpaul_ai/status/2113712534008262673