KernelCraft (PLENA): leaderboard
Metric: Kernel success rate (%) on the PLENA accelerator: functionally correct kernels within the iteration budget (15, 20 and 25 iterations for primitive, composite and end-to-end tasks) over the 100 task configurations that apply to it, read from the Total successful/total count; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 (Medium) | 56 | #105 (GPT-5.2) |
| 2 | Gemini 3 Flash (Preview) (Medium) | 35 | #78 (Gemini 3 Flash (Preview)) |
| 3 | Claude Sonnet 4 (20250514) (Thinking) | 12 | #211 (Claude Sonnet 4 (20250514)) |
| 4 | DeepSeek R1 0528 | 1 | #217 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=kernelcraft-plena · How It Works · Data refreshed daily, snapshot 2026-10-11.