# Language Model Alignment Evaluation on Arena-Hard v0.1

Source: [Language Model Alignment Evaluation on Arena-Hard v0.1](<https://www.sota2.com/research/sota/language-model-alignment-evaluation-on-arena-hard-v0-1>)

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

Data updated: 2026-02-26T06:08:56.785Z

Evaluation of language model alignment performance specifically on the Arena-Hard v0.1 dataset, typically measured by Win Rate percentage (WR %).

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 | Language Model Alignment Evaluation |
| Dataset (including any stated split) | Arena-Hard v0.1 |
| Metric | Win Rate (%) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 36 of 36 |
| 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

- [Win Rate (%)](<https://www.sota2.com/research/sota/language-model-alignment-evaluation-on-arena-hard-v0-1.md?metric=Win%20Rate%20(%25)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Win Rate (%) | Result date | Report paper |
| --- | --- | --- | --- |
| HyPO | 35.2 | Not recorded | 698eb041c97a0b3e3b5e682c |
| HyPO | 33.5 | Not recorded | 698eb041c97a0b3e3b5e682c |
| TR-DPO | 32.5 | Not recorded | 698eb041c97a0b3e3b5e682c |
| SimPO | 32.1 | Not recorded | 698eb041c97a0b3e3b5e682c |
| DPO | 31.5 | Not recorded | 698eb041c97a0b3e3b5e682c |
| RainbowPO | 31.3 | Not recorded | 698eb041c97a0b3e3b5e682c |
| TR-DPO | 31 | Not recorded | 698eb041c97a0b3e3b5e682c |
| KTO | 30.5 | Not recorded | 698eb041c97a0b3e3b5e682c |
| FocalPO | 30.2 | Not recorded | 698eb041c97a0b3e3b5e682c |
| SimPO | 30.1 | Not recorded | 698eb041c97a0b3e3b5e682c |
| CPO | 30 | Not recorded | 698eb041c97a0b3e3b5e682c |
| RainbowPO | 28.6 | Not recorded | 698eb041c97a0b3e3b5e682c |
| FocalPO | 27.9 | Not recorded | 698eb041c97a0b3e3b5e682c |
| HyPO | 25.2 | Not recorded | 698eb041c97a0b3e3b5e682c |
| SLiC-HF | 25.1 | Not recorded | 698eb041c97a0b3e3b5e682c |
| DPO | 23 | Not recorded | 698eb041c97a0b3e3b5e682c |
| CPO | 21 | Not recorded | 698eb041c97a0b3e3b5e682c |
| SimPO | 20.5 | Not recorded | 698eb041c97a0b3e3b5e682c |
| FocalPO | 18.7 | Not recorded | 698eb041c97a0b3e3b5e682c |
| KTO | 18.4 | Not recorded | 698eb041c97a0b3e3b5e682c |
| TR-DPO | 18.2 | Not recorded | 698eb041c97a0b3e3b5e682c |
| RainbowPO | 18.2 | Not recorded | 698eb041c97a0b3e3b5e682c |
| KTO | 17.5 | Not recorded | 698eb041c97a0b3e3b5e682c |
| SLiC-HF | 16.9 | Not recorded | 698eb041c97a0b3e3b5e682c |
| DPO | 15.4 | Not recorded | 698eb041c97a0b3e3b5e682c |
| SLiC-HF | 14.3 | Not recorded | 698eb041c97a0b3e3b5e682c |
| HyPO | 13.9 | Not recorded | 698eb041c97a0b3e3b5e682c |
| CPO | 12.2 | Not recorded | 698eb041c97a0b3e3b5e682c |
| SimPO | 11.2 | Not recorded | 698eb041c97a0b3e3b5e682c |
| TR-DPO | 9.5 | Not recorded | 698eb041c97a0b3e3b5e682c |
| RainbowPO | 9.2 | Not recorded | 698eb041c97a0b3e3b5e682c |
| FocalPO | 8.2 | Not recorded | 698eb041c97a0b3e3b5e682c |
| DPO | 7.9 | Not recorded | 698eb041c97a0b3e3b5e682c |
| KTO | 6.6 | Not recorded | 698eb041c97a0b3e3b5e682c |
| CPO | 6.4 | Not recorded | 698eb041c97a0b3e3b5e682c |
| SLiC-HF | 5.4 | Not recorded | 698eb041c97a0b3e3b5e682c |

### HyPO

Win Rate (%): 35.2. Result ID: 6998bc925698665ec41fadaa.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Instruct","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### HyPO

Win Rate (%): 33.5. Result ID: 6998bc925698665ec41fada1.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Base","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### TR-DPO

Win Rate (%): 32.5. Result ID: 6998bc925698665ec41fada8.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa771b1e209131cea7106a.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Instruct","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### SimPO

Win Rate (%): 32.1. Result ID: 6998bc925698665ec41fada6.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75901e209131cea68a5e.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Instruct","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### DPO

Win Rate (%): 31.5. Result ID: 6998bc925698665ec41fada3.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Instruct","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### RainbowPO

Win Rate (%): 31.3. Result ID: 6998bc925698665ec41fada9.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773f1e209131cea71f60.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Instruct","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### TR-DPO

Win Rate (%): 31. Result ID: 6998bc925698665ec41fad9f.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa771b1e209131cea7106a.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Base","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### KTO

Win Rate (%): 30.5. Result ID: 6998bc925698665ec41fada5.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75c61e209131cea69ebe.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Instruct","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### FocalPO

Win Rate (%): 30.2. Result ID: 6998bc925698665ec41fada7.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a411e209131cea81f52.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Instruct","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### SimPO

Win Rate (%): 30.1. Result ID: 6998bc925698665ec41fad9d.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75901e209131cea68a5e.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Base","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### CPO

Win Rate (%): 30. Result ID: 6998bc925698665ec41fada4.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73651e209131cea5c31e.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Instruct","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### RainbowPO

Win Rate (%): 28.6. Result ID: 6998bc925698665ec41fada0.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773f1e209131cea71f60.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Base","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### FocalPO

Win Rate (%): 27.9. Result ID: 6998bc925698665ec41fad9e.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a411e209131cea81f52.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Base","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### HyPO

Win Rate (%): 25.2. Result ID: 6998bc925698665ec41fad98.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Instruct","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### SLiC-HF

Win Rate (%): 25.1. Result ID: 6998bc925698665ec41fada2.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f521e209131cea4506a.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Instruct","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### DPO

Win Rate (%): 23. Result ID: 6998bc925698665ec41fad9a.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Base","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### CPO

Win Rate (%): 21. Result ID: 6998bc925698665ec41fad92.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73651e209131cea5c31e.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Instruct","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### SimPO

Win Rate (%): 20.5. Result ID: 6998bc925698665ec41fad94.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75901e209131cea68a5e.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Instruct","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### FocalPO

Win Rate (%): 18.7. Result ID: 6998bc925698665ec41fad95.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a411e209131cea81f52.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Instruct","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### KTO

Win Rate (%): 18.4. Result ID: 6998bc925698665ec41fad9c.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75c61e209131cea69ebe.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Base","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### TR-DPO

Win Rate (%): 18.2. Result ID: 6998bc925698665ec41fad96.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa771b1e209131cea7106a.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Instruct","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### RainbowPO

Win Rate (%): 18.2. Result ID: 6998bc925698665ec41fad97.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773f1e209131cea71f60.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Instruct","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### KTO

Win Rate (%): 17.5. Result ID: 6998bc925698665ec41fad93.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75c61e209131cea69ebe.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Instruct","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### SLiC-HF

Win Rate (%): 16.9. Result ID: 6998bc925698665ec41fad90.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f521e209131cea4506a.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Instruct","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### DPO

Win Rate (%): 15.4. Result ID: 6998bc925698665ec41fad91.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Instruct","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### SLiC-HF

Win Rate (%): 14.3. Result ID: 6998bc925698665ec41fad99.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f521e209131cea4506a.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Base","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### HyPO

Win Rate (%): 13.9. Result ID: 6998bc925698665ec41fad8f.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Base","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### CPO

Win Rate (%): 12.2. Result ID: 6998bc925698665ec41fad9b.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73651e209131cea5c31e.md>)

Recorded evaluation configuration: {"Model Family":"Llama-3","Model State":"Base","Model Parameters (B)":"8B"}. This record does not assert independent reproduction.

### SimPO

Win Rate (%): 11.2. Result ID: 6998bc925698665ec41fad8b.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75901e209131cea68a5e.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Base","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### TR-DPO

Win Rate (%): 9.5. Result ID: 6998bc925698665ec41fad8d.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa771b1e209131cea7106a.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Base","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### RainbowPO

Win Rate (%): 9.2. Result ID: 6998bc925698665ec41fad8e.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773f1e209131cea71f60.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Base","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### FocalPO

Win Rate (%): 8.2. Result ID: 6998bc925698665ec41fad8c.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a411e209131cea81f52.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Base","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### DPO

Win Rate (%): 7.9. Result ID: 6998bc925698665ec41fad88.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Base","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### KTO

Win Rate (%): 6.6. Result ID: 6998bc925698665ec41fad8a.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75c61e209131cea69ebe.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Base","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### CPO

Win Rate (%): 6.4. Result ID: 6998bc925698665ec41fad89.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73651e209131cea5c31e.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Base","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

### SLiC-HF

Win Rate (%): 5.4. Result ID: 6998bc925698665ec41fad87.

[Reporting paper](<https://www.sota2.com/research/paper/698eb041c97a0b3e3b5e682c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f521e209131cea4506a.md>)

Recorded evaluation configuration: {"Model Family":"Mistral","Model State":"Base","Model Parameters (B)":"7B"}. This record does not assert independent reproduction.

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