# b3: Backbone Breaker Benchmark (b3 Agent Security)

> Lakera's open-source benchmark of how well the LLM inside an AI agent resists attacks such as system-prompt exfiltration, malicious code injection and tool-call manipulation. Higher means more attacks resisted.

Canonical page: https://benchlm.ai/benchmarks/b3agentsecurity

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

## About b3 Agent Security

- Year: 2025
- Tasks: Threat snapshots of attacks on agent backends
- Format: Attack resistance rate
- Difficulty: Agent security
- Paper: [The Backbone Breaker Benchmark](https://www.lakera.ai/blog/the-backbone-breaker-benchmark)

Mistral reported these results in its October 29, 2026 Mistral Large 4 open-weight post, including the rival-model values in its chart. They are provider-run, so BenchLM shows them display-only and keeps them out of weighted scores. Rival values were published as whole percentages.

b3 Agent Security is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (4 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Mistral Large 4](/models/mistral-large-4) | Mistral | 97.1% |
| 2 | [GLM-5.3](/models/glm-5-3) | Z.AI | 93.0% |
| 3 | [Kimi K3](/models/kimi-k3) | Moonshot AI | 88.0% |
| 4 | [Gemma 4 31B](/models/gemma-4-31b) | Google | 83.0% |

## FAQ

### What does b3 Agent Security measure?

Lakera's open-source benchmark of how well the LLM inside an AI agent resists attacks such as system-prompt exfiltration, malicious code injection and tool-call manipulation. Higher means more attacks resisted.

### Which model scores highest on b3 Agent Security?

Mistral Large 4 by Mistral currently leads with a score of 97.1% on b3 Agent Security.

### How many models are evaluated on b3 Agent Security?

4 AI models have published results on b3 Agent Security in the BenchLM catalog.

### Does b3 Agent Security affect BenchLM's overall score?

Not directly. b3 Agent Security is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Compare Top Models on b3 Agent Security

- [Mistral Large 4 vs GLM-5.3](/compare/glm-5-3-vs-mistral-large-4)
- [GLM-5.3 vs Kimi K3](/compare/glm-5-3-vs-kimi-k3)
- [Kimi K3 vs Gemma 4 31B](/compare/gemma-4-31b-vs-kimi-k3)