Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
external benchmark
ExploitBench v8-bench. Every model the catalog carries a published ExploitBench value for, ranked by that value.
A cybersecurity benchmark for evaluating LLM agents on full-control V8 exploit synthesis using 16 measured exploit capability flags. 18 models in the ModelChoice catalog have a published value.
Each point is a model with a published ExploitBench value and complete token prices, on the 75% input / 25% output blend. Frontier models are spaced evenly along the cost axis.
18 models with a published ExploitBench 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 | Capability coverage percentage over 16 flags |
|---|---|---|---|
| 1 | OpenAI | 100 | |
| 1 | OpenAI | 100 | |
| 1 | OpenAI | 100 |
What the benchmark measures and how its value is published, from the source’s own record.
ExploitBench (ExploitBench v8-bench) is an external benchmark published by ExploitBench. A cybersecurity benchmark for evaluating LLM agents on full-control V8 exploit synthesis using 16 measured exploit capability flags. Its evaluation set is described by the source as v8 exploit synthesis runs. The source rates its difficulty as browser exploitation and cybersecurity.
Each model's result is published by ExploitBench as capability coverage percentage over 16 flags. ModelChoice prints that value to 0 decimal places, exactly as published, and never converts or rescales it. This page ranks the 18 models that carry a published value, highest value first; a model the source never scored is left out rather than ranked at zero. The source does not state whether a higher value is the better result, so neither does this page.
| 1 | OpenAI | 100 |
| 5 | OpenAI | 100 |
| 6 | OpenAI | 97 |
| 7 | Anthropic | 95 |
| 8 | Anthropic | 78 |
| 9 | Moonshot AI | 75 |
| 10 | OpenAI | 74 |
| 11 | Anthropic | 69 |
| 12 | Z.AI | 54 |
| 13 | OpenAI | 53 |
| 14 | Xiaomi | 48 |
| 15 | Cognition | 35 |
| 16 | OpenAI | 33 |
| 17 | Moonshot AI | 32 |
| 18 | Xiaomi | 25 |
Evidence key: Observed
Rows are ordered by the value ExploitBench published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.