Gray Swan IPI — Combined Attack Set, Transfer Only, No Computer Use: leaderboard
Gray Swan IPI evaluates indirect prompt-injection robustness in text-only settings using a combined attack set in transfer-only mode, without computer use. The model is tested on whether it resists attacks, and the pinned metric is attack success rate at 15 attempts; lower values indicate fewer successful attacks. Source: BenchmarkList’s attributed result table.
Metric: Attack success rate @15 (%). Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Carbon | 0 | #1 |
| 2 | Gemini 4 Argon | 0.7 | #9 |
| 3 | Claude Opus 5.5 | 1 | #3 |
| 4 | Claude Fable 5.1 | 1 | #10 |
| 5 | Claude Opus 5 | 4.6 | #16 |
| 6 | Gemini 3.8 Flash | 5.5 | #30 |
| 7 | GPT-6 Astra | 8.5 | #6 |
| 8 | GPT-6.1 Sol | 8.6 | #7 |
| 9 | Muse Spark 1.3 | 15.9 | #31 |
| 10 | GLM-5.3 | 31.5 | #44 |
| 11 | Grok 4.6 | 51.8 | #29 |
| 12 | Kimi K3 | 52.7 | #35 |
Interactive version: theaggregate.ai/benchmark?slug=gray-swan-ipi-combined-attack-set-transfer-only-no-computer-use · How It Works · Data refreshed daily, snapshot 2026-11-05.