Coding benchmark
FrontierCode 1.1 Extended leaderboard
Every model the catalog carries a published FrontierCode 1.1 Extended value for, ranked by that value.
CategoryCoding
MeasureRepository task completion with maintainer rubrics
Tasks150 private software-engineering tasks
DifficultyFrontier coding-agent quality
Published byGPT-5.6 models are now available in Devin
Cognition's 150-task Extended subset of the FrontierCode 1.1 software-engineering benchmark.
FrontierCode 1.1 Extended ranking
6 models with a published FrontierCode 1.1 Extended 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 |  GPT-6 Astra | OpenAI | 64.5 |
| 2 |  Claude Opus 5 | Anthropic | 63.6 |
| 3 |  Grok 4.6 | xAI | 61.3 |
| 4 | GPT-5.6 Sol | OpenAI | 60.6 |
| 5 | GPT-5.6 Terra | OpenAI | 55.8 |
| 6 | GPT-5.6 Luna | OpenAI | 55.1 |
Evidence key: Observed
Rows are ordered by the value GPT-5.6 models are now available in Devin 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.