# IHEval leaderboard

Instruction Following benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelscale.dev/leaderboards/benchmarks/instruction-following-ih-eval

IHEval: instruction hierarchy evaluation. Every model the catalog carries a published IHEval value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Instruction Following |
| Measure | Conflict-setting accuracy |
| Tasks | Aligned and conflicting instruction sets |
| Difficulty | Instruction priority |
| Published by | [IHEval: Evaluating Language Models on Following the Instruction Hierarchy](https://github.com/ytyz1307zzh/IHEval) |

Tests whether a model follows the priority order of system messages, user messages, conversation history and tool outputs, including when their instructions conflict.

## IHEval ranking

2 models with a published IHEval value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Conflict-setting accuracy | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Nemotron 3.5 Bolt E4B BF16](https://modelscale.dev/models/nemotron-3-5-bolt-e4b-bf16) | NVIDIA | 84.79 | Observed 2026-11-21 · source benchlm:models |
| 2 | [Nemotron 3.5 Bolt E4B](https://modelscale.dev/models/nemotron-3-5-bolt-e4b) | NVIDIA | 82.40 | Observed 2026-11-21 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
