SAKE - Five-Shot - Architectural Solutions: leaderboard
Metric: Accuracy (%). Source: arxiv.org. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 90.6 |
| 2 | Claude Sonnet 4.6 | 89.8 |
| 3 | GPT-5.4 | 89.8 |
| 4 | Gemini 3.1 Flash Lite | 89.8 |
| 5 | Grok 4.20 | 89.4 |
| 6 | Mistral Small 4 | 89 |
| 7 | Qwen 3 235B A22B | 88.6 |
| 8 | Claude Haiku 4.5 | 88.2 |
| 9 | Grok 4.1 Fast | 87.8 |
| 10 | DeepSeek V3.2 | 85.4 |
| 11 | GPT-5.4 Nano | 83.7 |
Interactive version: theaggregate.ai/benchmark?slug=sake-five-shot-architectural-solutions · How It Works · Data refreshed daily, snapshot 2026-09-19.