RULER — Rollup bucket mixing RULER (128K)/(1M) raws across many models — not a single-length benchmark.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | nvidia-nemotron-3-ultra-550b-a55b | NVIDIA | 94.7 | 1 | 2026-08-24 |
| 2 | Nemotron 3 Ultra 550B A55B | NVIDIA | 94.7 | 1 | 2026-06-10 |
| 3 | Seed Oss 36B Instruct | ByteDance | 94.6 | 1 | 2026-06-15 |
| 4 | Qwen3 30B A3B | Alibaba | 94.5 | 1 | 2026-06-15 |
| 5 | Nemotron 3 Super 120B A12B | NVIDIA | 91.8 | 1 | 2026-08-23 |
| 6 | Phi 3.5 MoE Instruct | Microsoft | 87.1 | 1 | 2026-08-23 |
| 7 | Phi 3.5 Mini Instruct | Microsoft | 84.1 | 1 | 2026-08-23 |
| 8 | Qwen3 32B | Alibaba | 77.5 | 1 | 2026-06-15 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.