# Cybench 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-cybench

Every model the catalog carries a published Cybench value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Agentic |
| Measure | Cybersecurity agent task completion |
| Tasks | 40 professional CTF tasks |
| Difficulty | Professional cybersecurity |
| Published by | [Cybench: A Framework for Evaluating Cybersecurity Capabilities and Risks of Language Models](https://arxiv.org/abs/2408.08926) |

A cybersecurity benchmark of professional Capture the Flag tasks for measuring autonomous cyber agent capability and risk.

## Cybench ranking

2 models with a published Cybench 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 | Cybersecurity agent task completion | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Mistral Large 4](https://modelchoice.dev/models/mistral-large-4) | Mistral | 93.0 | Observed 2026-11-21 · source benchlm:models |
| 2 | [Muse Spark 1.1](https://modelchoice.dev/models/muse-spark-1-1) | Meta | 92.9 | 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.
