Anthropic AI R&D - Kernel Task Best Speedup: leaderboard
Kernel optimization task from Anthropic's internal AI R&D rule-out suite: best speedup (x) achieved on the hard task with the standard scaffold; unbounded, higher is better. From Table 2.3.5.A of the Claude Opus 5 System Card (July 2026); the Sonnet 5 row is from Table 2.3.3.A of the Sonnet 5 card.
Metric: Best Speedup (x). Source: www-cdn.anthropic.com. Saturation forecast: Around February 2028. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 | 449.46 | #16 |
| 2 | Claude Mythos 5 | 430.93 | #8 |
| 3 | Claude Opus 4.7 | 371.75 | #51 |
| 4 | Claude Sonnet 5 | 284.52 | #65 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=anthropic-ai-r-d-kernel-task-best-speedup · How It Works · Data refreshed daily, snapshot 2026-11-05.