SWE-PI

Only in system cards

A coding agent does a bug-fix task in a repository seeded with an injected payload that tries to exfiltrate a secret. Reports attack success rate.

In-house Meta benchmark. 142 scenarios from 50 SWE-bench instances.

Made by: Meta
Kind: Secure code and review
Level: Expert
Scored: Checked automatically
Answers: Agent task
Tasks: Kept private

Strongest model tested: Muse Spark 1.3 #26 on the capability index
Last in a system card: October 2026
Last model added: October 2026

Model scored in lab cards

System cards report it, from 1 lab

Muse Spark 1.3 leads with 0%

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.

Bars Table

1 Muse Spark 1.3 0%

Source: Labs’ model cards

#	Model	Score	Setting	Date	Source
1	Muse Spark 1.3	0%		Oct 2026	Lab model card

Scores over time

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

Reported by the lab in its model card

Source: 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.

2 system cards report it, from 1 lab

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
Oct 23, 2026	Muse Spark 1.3 system card	0	SWE-PI pass@1 ASR (secret exfiltration via prompt injection), Muse Spark 1.3 (API) (Figure 26)	Robustness
Jul 9, 2026	Muse Spark 1.1 system card		SWE-PI Synthetic and SWE-PI Agent subsets, no overall (Figure 29, 30). Synthetic pass@1 ASR 0.0	Robustness

Muse Spark 1.3 leads with 0%

Scores over time

Labs report it only in system cards

2 system cards report it, from 1 lab

More in Cybersecurity