# FrontierCode

14 Hacker News commenters recommended FrontierCode and 4 criticized it, November 2025 to November 2026. It was recommended in 5 of 13 months. It meets the Hacker Approved badge rule.

**Verdict:** Recommended as a public coding benchmark that grades whether code would get merged and was calibrated with open source maintainers. Critics question its scores and results.

## Why people recommend it

- Well reasoned and carefully built (3 comments): "they are very well defined and solid coverage of everything one should want out of code" ([source](https://news.ycombinator.com/item?id=48453609)) "Well reasoned, tons of work put into eval, thanks for building it." ([source](https://news.ycombinator.com/item?id=48453013)) "more thoughtful than any benchmark that I've seen until now" ([source](https://news.ycombinator.com/item?id=48454489))
- Top pick among coding benchmarks (3 comments): "DeepSWE and FrontierCode are more realistic if you read up on what they actually measure." ([source](https://news.ycombinator.com/item?id=48994637)) "Cognition's Frontier Code is probably the best large public benchmark at this." ([source](https://news.ycombinator.com/item?id=49030780)) "Only look at DeepSWE and FrontierCode right now for coding imo." ([source](https://news.ycombinator.com/item?id=48849363))
- Grades whether code would get merged (2 comments): "The later is more interesting for me personally, due to the design of the benchmark heavily grading "mergability"" ([source](https://news.ycombinator.com/item?id=48837848)) "First benchmark to measure: "would this code get actually merged?"" ([source](https://news.ycombinator.com/item?id=48452470))
- Matches real-world coding experience (2 comments): "the only benchmark that aligns with my actual LLM experiences" ([source](https://news.ycombinator.com/item?id=49556382)) "FrontierCode matches the vibes a lot better" ([source](https://news.ycombinator.com/item?id=48763727))

## What people warn about

- Maker's own benchmark, scores questioned (2 comments): "AA seems to have accepted Cognition's benchmark scores without questioning" ([source](https://news.ycombinator.com/item?id=50269895)) "on their own benchmark that they made" ([source](https://news.ycombinator.com/item?id=49649332))
- Results contradict other benchmarks (2 comments): "Any benchmark that shows gpt-6 sol as better than gpt-5.6 sol is incredibly suspect." ([source](https://news.ycombinator.com/item?id=50271700)) "this has dramatically different results, I don't trust it at all!" ([source](https://news.ycombinator.com/item?id=50266542))

## Cite as

Hacker Approved, "FrontierCode", https://hackerapproved.com/p/frontiercode/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).