# InverseIFEval

> Tests whether a model follows counterintuitive instructions that run against conventions it learned in training.

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

- Category: [Instruction Following](/instruction-following)
- Last updated: October 29, 2026

## About InverseIFEval

- Year: 2025
- Tasks: Counterintuitive instructions
- Format: Overall accuracy
- Difficulty: Overriding trained conventions
- Paper: [InverseIFEval](https://arxiv.org/abs/2509.04292)

NVIDIA reports it for Nemotron 3.5 Bolt E4B (October 23, 2026), run in its NeMo Gym harness, as overall accuracy in reasoning mode.

InverseIFEval 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 | 46.84% |
| 2 | [Nemotron 3.5 Bolt E4B](/models/nemotron-3-5-bolt-e4b) | NVIDIA | 45.26% |

## FAQ

### What does InverseIFEval measure?

Tests whether a model follows counterintuitive instructions that run against conventions it learned in training.

### Which model scores highest on InverseIFEval?

Nemotron 3.5 Bolt E4B BF16 by NVIDIA currently leads with a score of 46.84% on InverseIFEval.

### How many models are evaluated on InverseIFEval?

2 AI models have published results on InverseIFEval in the BenchLM catalog.

### Does InverseIFEval affect BenchLM's overall score?

Not directly. InverseIFEval is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Compare Top Models on InverseIFEval

- [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)