# The benchmarks

> 6 benchmarks killed by being solved.

- [GLUE](https://models.rip/glue/epitaph) — 2018-04-20 — 2019-07-12 · frontier 70% → 88.4%
- [SuperGLUE](https://models.rip/superglue/epitaph) — 2019-05-02 — 2021-01-06 · frontier 69% → 90.3%
- [HellaSwag](https://models.rip/hellaswag/epitaph) — 2019-05-19 — 2023-03-14 · frontier 47.3% → 95.3%
- [MMLU](https://models.rip/mmlu/epitaph) — 2020-09-07 — 2025-02-24 · frontier 43.9% → 91.8%
- [HumanEval](https://models.rip/humaneval/epitaph) — 2021-07-07 — 2025-02-24 · frontier 28.8% → 92.7%
- [GSM8K](https://models.rip/gsm8k/epitaph) — 2021-10-27 — 2024-05-01 · frontier 55% → 91.1%

## Elsewhere

- [The whole cemetery](https://models.rip/)
- [The dataset](https://models.rip/graveyard.json)