# MuSR leaderboard

Reasoning 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/reasoning-musr

Testing the Limits of Chain-of-thought with Multistep Soft Reasoning. Every model the catalog carries a published MuSR value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Reasoning |
| Measure | Narrative-based reasoning |
| Tasks | Multi-step reasoning |
| Difficulty | Complex reasoning tasks |
| Published by | [MuSR: Testing the Limits of Chain-of-thought with Multistep Soft Reasoning](https://arxiv.org/abs/2310.16049) |

A dataset for evaluating language models on multistep soft reasoning tasks specified in natural language narratives. Tests the ability to perform complex, structured reasoning.

## MuSR ranking

6 models with a published MuSR 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 | Narrative-based reasoning | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Ternary Bonsai 8B](https://modelscale.dev/models/ternary-bonsai-8b) | Prism ML | 56.2 | Observed 2026-11-21 · source benchlm:models |
| 2 | [Ternary Bonsai 1.7B](https://modelscale.dev/models/ternary-bonsai-1-7b) | Prism ML | 50.8 | Observed 2026-11-21 · source benchlm:models |
| 3 | [1-bit Bonsai 8B](https://modelscale.dev/models/bonsai-8b) | Prism ML | 50 | Observed 2026-11-21 · source benchlm:models |
| 4 | [1-bit Bonsai 1.7B](https://modelscale.dev/models/bonsai-1-7b) | Prism ML | 45.1 | Observed 2026-11-21 · source benchlm:models |
| 4 | [Ternary Bonsai 4B](https://modelscale.dev/models/ternary-bonsai-4b) | Prism ML | 45.1 | Observed 2026-11-21 · source benchlm:models |
| 6 | [1-bit Bonsai 4B](https://modelscale.dev/models/bonsai-4b) | Prism ML | 41.4 | 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.
