# b3 Agent Security leaderboard

Agentic benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelchoice.dev/leaderboards/benchmarks/agentic-b3-agent-security

b3: Backbone Breaker Benchmark. Every model the catalog carries a published b3 Agent Security value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Agentic |
| Measure | Attack resistance rate |
| Tasks | Threat snapshots of attacks on agent backends |
| Difficulty | Agent security |
| Published by | [The Backbone Breaker Benchmark](https://www.lakera.ai/blog/the-backbone-breaker-benchmark) |

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.

## b3 Agent Security ranking

4 models with a published b3 Agent Security value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Attack resistance rate | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Mistral Large 4](https://modelchoice.dev/models/mistral-large-4) | Mistral | 97.1 | Observed 2026-11-21 · source benchlm:models |
| 2 | [GLM-5.3](https://modelchoice.dev/models/glm-5-3) | Z.AI | 93.0 | Observed 2026-11-21 · source benchlm:models |
| 3 | [Kimi K3](https://modelchoice.dev/models/kimi-3) | Moonshot AI | 88.0 | Observed 2026-11-21 · source benchlm:models |
| 4 | [Gemma 4 31B](https://modelchoice.dev/models/gemma-4-31b) | Google | 83.0 | Observed 2026-11-21 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
