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
PMPP-Hard GPU kernel optimization (AUC, 0-10 hours). Every model the catalog carries a published PMPP-Hard value for, ranked by that value.
An agentic GPU kernel-optimization benchmark of 69 hard tasks, adapted by Thinking Machines Lab for consistent cross-model evaluation. Each model gets four independent attempts with a 10-hour budget, and the score is the area under the mean tasks-solved curve, out of a maximum of 690 task-hours. 2 models in the ModelChoice catalog have a published value.
Each point is a model with a published PMPP-Hard value and complete token prices, on the 75% input / 25% output blend. Frontier models are spaced evenly along the cost axis.
2 models with a published PMPP-Hard 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 | Area under the mean tasks-solved curve over 10 hours (task-hours, max 690) |
|---|---|---|---|
| 1 | TMInkling 2 Small | Thinking Machines Lab | 494.27 |
| 2 | TMInkling 2 Mini | Thinking Machines Lab | 455.82 |
Evidence key: Observed
Rows are ordered by the value Introducing Inkling 2 Small 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.
What the benchmark measures and how its value is published, from the source’s own record.
PMPP-Hard (PMPP-Hard GPU kernel optimization (AUC, 0-10 hours)) is an external benchmark published by Introducing Inkling 2 Small. An agentic GPU kernel-optimization benchmark of 69 hard tasks, adapted by Thinking Machines Lab for consistent cross-model evaluation. Each model gets four independent attempts with a 10-hour budget, and the score is the area under the mean tasks-solved curve, out of a maximum of 690 task-hours. Its evaluation set is described by the source as 69 hard GPU kernel-optimization tasks. The source rates its difficulty as long-horizon GPU kernel optimization.
Each model's result is published by Introducing Inkling 2 Small as area under the mean tasks-solved curve over 10 hours (task-hours, max 690). ModelChoice prints that value to 2 decimal places, exactly as published, and never converts or rescales it. This page ranks the 2 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.