# RULER at a 64K-token context (RULER (64K))

> Synthetic long-context benchmark of retrieval, multi-hop tracing, aggregation and question answering, scored at a 64K-token context.

Canonical page: https://benchlm.ai/benchmarks/ruler64k

- Category: [Reasoning](/reasoning)
- Last updated: October 29, 2026

## About RULER (64K)

- Year: 2024
- Tasks: Synthetic long-context tasks
- Format: Overall accuracy at 64K
- Difficulty: Long-context retrieval and reasoning
- Paper: [RULER: What's the Real Context Size of Your Long-Context Language Models?](https://github.com/NVIDIA/RULER)

NVIDIA reports it for Nemotron 3.5 Bolt E4B (October 23, 2026), run in its NeMo Gym harness, as overall accuracy at 64K with reasoning off. The RULER context-length report compares 128K values, so 64K launch reports stay on this key.

RULER (64K) is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (2 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Nemotron 3.5 Bolt E4B BF16](/models/nemotron-3-5-bolt-e4b-bf16) | NVIDIA | 86.99% |
| 2 | [Nemotron 3.5 Bolt E4B](/models/nemotron-3-5-bolt-e4b) | NVIDIA | 85.76% |

## FAQ

### What does RULER (64K) measure?

Synthetic long-context benchmark of retrieval, multi-hop tracing, aggregation and question answering, scored at a 64K-token context.

### Which model scores highest on RULER (64K)?

Nemotron 3.5 Bolt E4B BF16 by NVIDIA currently leads with a score of 86.99% on RULER (64K).

### How many models are evaluated on RULER (64K)?

2 AI models have published results on RULER (64K) in the BenchLM catalog.

### Does RULER (64K) affect BenchLM's overall score?

Not directly. RULER (64K) is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Compare Top Models on RULER (64K)

- [Nemotron 3.5 Bolt E4B BF16 vs Nemotron 3.5 Bolt E4B](/compare/nemotron-3-5-bolt-e4b-vs-nemotron-3-5-bolt-e4b-bf16)