# Model Inertia measurement on MFQ-30

Source: [Model Inertia measurement on MFQ-30](<https://www.sota2.com/research/sota/model-inertia-measurement-on-mfq-30>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-21T10:45:46.654Z

Evaluation of Model Inertia measurement performance on the MFQ-30 dataset, using metrics such as Harm, Fair, Ingr, Auth, Pur, and Avg.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Model Inertia measurement |
| Dataset (including any stated split) | MFQ-30 |
| Metric | Harm |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Harm](<https://www.sota2.com/research/sota/model-inertia-measurement-on-mfq-30.md?metric=Harm>): Higher is better; unit: PERCENT.

- [Fair](<https://www.sota2.com/research/sota/model-inertia-measurement-on-mfq-30.md?metric=Fair>): Higher is better; unit: PERCENT.

- [Ingr](<https://www.sota2.com/research/sota/model-inertia-measurement-on-mfq-30.md?metric=Ingr>): Higher is better; unit: PERCENT.

- [Auth](<https://www.sota2.com/research/sota/model-inertia-measurement-on-mfq-30.md?metric=Auth>): Higher is better; unit: PERCENT.

- [Pur](<https://www.sota2.com/research/sota/model-inertia-measurement-on-mfq-30.md?metric=Pur>): Higher is better; unit: PERCENT.

- [Avg](<https://www.sota2.com/research/sota/model-inertia-measurement-on-mfq-30.md?metric=Avg>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Harm | Result date | Report paper |
| --- | --- | --- | --- |
| LLaMA 3 70B | 71.9 | Not recorded | 68fa77dc1e209131cea759fe |
| GPT-4o | 58.9 | Not recorded | 68fa77dc1e209131cea759fe |
| Claude 3 Haiku | 51.6 | Not recorded | 68fa77dc1e209131cea759fe |
| Claude 3 Sonnet | 48.5 | Not recorded | 68fa77dc1e209131cea759fe |
| GPT-3.5 Turbo | 35.5 | Not recorded | 68fa77dc1e209131cea759fe |
| Claude 3 Opus | 29.9 | Not recorded | 68fa77dc1e209131cea759fe |
| LLaMA 3 8B | 26 | Not recorded | 68fa77dc1e209131cea759fe |

### LLaMA 3 70B

Harm: 71.9. Result ID: 69e753a747f34ce924ce52b1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77dc1e209131cea759fe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"Seeds":"3"}. This record does not assert independent reproduction.

### GPT-4o

Harm: 58.9. Result ID: 69e753a747f34ce924ce52b2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77dc1e209131cea759fe.md>)

Recorded evaluation configuration: {"Seeds":"3"}. This record does not assert independent reproduction.

### Claude 3 Haiku

Harm: 51.6. Result ID: 69e753a747f34ce924ce52b4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77dc1e209131cea759fe.md>)

Recorded evaluation configuration: {"Seeds":"3"}. This record does not assert independent reproduction.

### Claude 3 Sonnet

Harm: 48.5. Result ID: 69e753a747f34ce924ce52b3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77dc1e209131cea759fe.md>)

Recorded evaluation configuration: {"Seeds":"3"}. This record does not assert independent reproduction.

### GPT-3.5 Turbo

Harm: 35.5. Result ID: 69e753a747f34ce924ce52b6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77dc1e209131cea759fe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa727a1e209131cea56a1e.md>)

[Implementation](<https://github.com/openai/evals>)

Recorded evaluation configuration: {"Seeds":"3"}. This record does not assert independent reproduction.

### Claude 3 Opus

Harm: 29.9. Result ID: 69e753a747f34ce924ce52b5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77dc1e209131cea759fe.md>)

Recorded evaluation configuration: {"Seeds":"3"}. This record does not assert independent reproduction.

### LLaMA 3 8B

Harm: 26. Result ID: 69e753a747f34ce924ce52b7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77dc1e209131cea759fe.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"Seeds":"3"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.