CyberSecEval Instruct

Muse Spark 1.1 77.3% leads since July 2026

Code Llama 34B Instruct 62.5% leads

Muse Spark 1.1 77.3% leads since July 2026

Only in system cards

Labs’ own numbers All independent runs Include claimed scores

Models get coding requests built to elicit known insecure patterns. The Insecure Code Detector flags vulnerable code, and the score is the pass rate.

Part of every CyberSecEval release since version 1. Some reports give the vulnerable rate, which equals 100 minus the pass rate.

Made by: Meta (PurpleLlama)
Released: December 2023
Kind: Secure code and review
Level: University
Scored: Checked automatically
Answers: Code
Tasks: All public

Strongest model tested: Muse Spark 1.1 #47 on the capability index
Last in a system card: July 2026
Last model added: July 2026
Scores refreshed: November 12th, 2026

In the authors’ words

This paper presents CyberSecEval, a comprehensive benchmark developed to help bolster the cybersecurity of Large Language Models (LLMs) employed as coding assistants. As what we believe to be the most extensive unified cybersecurity safety benchmark to date, CyberSecEval provides a thorough evaluation of LLMs in two crucial security domains: their propensity to generate insecure code and their level of compliance when asked to assist in cyberattacks. Through a case study involving seven models from the Llama 2, Code Llama, and OpenAI GPT large language model families, CyberSecEval effectively pinpointed key cybersecurity risks. More importantly, it offered practical insights for refining these models. A significant observation from the study was the tendency of more advanced models to suggest insecure code, highlighting the critical need for integrating security considerations in the development of sophisticated LLMs. CyberSecEval, with its automated test case generation and evaluation pipeline covers a broad scope and equips LLM designers and researchers with a tool to broadly measure and enhance the cybersecurity safety properties of LLMs, contributing to the development of more secure AI systems.

Abstract of Purple Llama CyberSecEval: A Secure Coding Benchmark for Language Models (arXiv:2312.04724, CC0)

Models scored in lab cards

System cards report it, from 2 labs

Muse Spark 1.1 leads with 77.3% among models released by July 2026

Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.

Code Llama 34B Instruct leads with 62.5% across all independent runs

Best score per model across the official leaderboard, Epoch AI’s own runs and any board Epoch AI copied. Each bar and row names its source.

Muse Spark 1.1 leads with 77.3%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model.

Bars Table

The newest model with a score here came out in July 2026, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

Source: Labs’ model cards

62.5 Code Llama 34B Instruct

1 Code Llama 34B Instruct Instruct 62.5%

Source: Official CyberSecEval Instruct leaderboard

1 Muse Spark 1.1 77.3%
2 Muse Spark 69.5%
3 MAI-Thinking-1 63%
4 Code Llama 34B Instruct Instruct 62.5%

Sources: Labs’ model cards and official CyberSecEval Instruct leaderboard

#	Model	Score	Date	Source
1	Muse Spark 1.1	77.3%	Jul 2026	Lab model card
2	Muse Spark	69.5%	Jun 2026	Lab model card
3	MAI-Thinking-1	63%	Jun 2026	Lab model card

#	Model	Score	Setting	Date	Source
1	Code Llama 34B Instruct	62.5%	Instruct	Aug 2023	Official leaderboard

#	Model	Score	Setting	Date	Source
1	Muse Spark 1.1	77.3%		Jul 2026	Lab model card
2	Muse Spark	69.5%		Jun 2026	Lab model card
3	MAI-Thinking-1	63%		Jun 2026	Lab model card
4	Code Llama 34B Instruct	62.5%	Instruct	Aug 2023	Official leaderboard

The benchmark's own leaderboard has 4 entries, 1 of them models we track. Switch the scores above to rank them together.

3 models have held number one in labs’ own numbers since its release in December 2023 Number one over time 4 models have held number one with claims included since its release in December 2023

Lab claims count from the day the card came out, independent runs from the model’s release.

Microsoft Meta

Lab claims Not tested since Jul 2026

Muse Spark 1.1 77.3%

Source: Labs’ model cards

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Reported by the lab in its model card Best so far

Sources: Official CyberSecEval Instruct leaderboard and labs’ model cards

Labs report it only in system cards

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark. Some system cards do, below.

3 system cards report it, from 2 labs

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

Date	Document	Score	As the document writes it	Section
Jul 9, 2026	Muse Spark 1.1 system card	77.3	CyberSecEval instruct, pass@5 secure-code rate (Figure 13)	Cyber risk
Jun 12, 2026	Muse Spark system card	69.5	CyberSecEval instruct secure-code rate (2.2.4 text)	Cyber risk
Jun 2, 2026	MAI-Thinking-1 technical report	63	CyberSec Instruct, CyberSecEval 4 insecure code generation, higher is better (Table 12, rounded)	Cyber risk