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.
Coding benchmark
Every model the catalog carries a published FrontierCode 1.1 Main value for, ranked by that value.
Cognition's 100-task software-engineering benchmark for whether coding agents produce mergeable, production-quality pull requests, scored for correctness, tests, scope, style, and maintainability through maintainer-authored rubrics. 17 models in the ModelScale catalog have a published value.
Each point is a model with a published FrontierCode 1.1 Main value and complete token prices, on the 75% input / 25% output blend. Frontier models are spaced evenly along the cost axis.
17 models with a published FrontierCode 1.1 Main 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 | Repository task completion with maintainer rubrics |
|---|---|---|---|
| 1 | Anthropic | 54.4 | |
| 2 | Anthropic | 53.5 | |
| 3 | Anthropic | 53.4 |
What the benchmark measures and how its value is published, from the source’s own record.
FrontierCode 1.1 Main is a coding benchmark published by FrontierCode leaderboard. Cognition's 100-task software-engineering benchmark for whether coding agents produce mergeable, production-quality pull requests, scored for correctness, tests, scope, style, and maintainability through maintainer-authored rubrics. Its evaluation set is described by the source as 100 private Main tasks (150 in Extended). The source rates its difficulty as frontier coding-agent quality.
Each model's result is published by FrontierCode leaderboard as repository task completion with maintainer rubrics. ModelScale prints that value to 1 decimal place, exactly as published, and never converts or rescales it. This page ranks the 17 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.
| 4 | OpenAI | 53.3 |
| 5 | Cognition | 50.0 |
| 6 | Anthropic | 46.5 |
| 7 | Anthropic | 46.4 |
| 8 | Anthropic | 46.2 |
| 9 | Cognition | 45.5 |
| 10 | 43.6 |
| 11 | OpenAI | 43.0 |
| 12 | Anthropic | 42.7 |
| 13 | Cognition | 42.3 |
| 14 | Anthropic | 38.5 |
| 15 | OpenAI | 27.0 |
| 16 | Anthropic | 26.9 |
| 17 | Anthropic | 24.3 |
Evidence key: Observed
Rows are ordered by the value FrontierCode leaderboard 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.