Reward Modeling on RewardBench v2 (test)
86.5Average ScoreSkywork-Reward-V2-Llama-3.1-8B-40M
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Skywork-Reward-V2-Llama-3.1-8B-40MBackbone=Llama-3.1, Parameters=8B, Curated Data Volume=40M2025.07 | 86.5 | 87.9 | 67.8 | 83.1 | 97.3 | 99.2 | 83.9 | |
| Skywork-Reward-V2-Llama-3.1-8B-40MModel Type=Reward Model, Backbone=Llama-3.1, Parameter Count=8B, Training Data=40M2025.07 | 86.5 | — | — | — | — | — | — | |
| Skywork-Reward-V2-Llama-3.1-8BBackbone=Llama-3.1, Parameters=8B2025.07 | 84.1 | 84.6 | 66.2 | 77.6 | 96.7 | 98.4 | 81.2 | |
| Skywork-Reward-V2-Llama-3.1-8BModel Type=Reward Model, Backbone=Llama-3.1, Parameter Count=8B2025.07 | 84.1 | — | — | — | — | — | — | |
| LMUnit-qwen2.5-72bBackbone=Qwen-2.5-72B2025.07 | 82.1 | 87.2 | 54.4 | 72.7 | 91.3 | 96.8 | 90.1 | |
| LMUnit-llama3.1-70bBackbone=Llama-3.1-70B2025.07 | 80.5 | 84.6 | 48.8 | 71.6 | 90.7 | 97 | 90.6 | |
| Skywork-Reward-V2-Qwen3-8BBackbone=Qwen3, Parameters=8B2025.07 | 78.2 | 79.8 | 49.1 | 77 | 94 | 96.4 | 72.9 | |
| Skywork-Reward-V2-Qwen3-8BModel Type=Reward Model, Backbone=Qwen3, Parameter Count=8B2025.07 | 78.2 | — | — | — | — | — | — | |
| gemini-2.5-flash-preview-04-17Release Date=04-172025.07 | 77.2 | 65.7 | 55.3 | 81.1 | 90.9 | 86.7 | 83.4 | |
| QRM-Gemma-2-27BBackbone=Gemma-2-27B2025.07 | 76.7 | 78.5 | 37.2 | 69.9 | 95.8 | 95.4 | 83.2 | |
| Llama-3.1-Nemotron-70BModel Type=Open Reward Model, Backbone=Llama-3.1, Parameter Count=70B2025.07 | 76.7 | — | — | — | — | — | — | |
| INF-ORM-Llama3.1-70BBackbone=Llama-3.1-70B2025.07 | 76.5 | 74.1 | 41.9 | 69.9 | 96.4 | 90.3 | 86.2 | |
| claude-opus-4-20250514Release Date=202505142025.07 | 76.5 | 82.7 | 41.9 | 74.9 | 89.5 | 86.2 | 83.7 | |
| INF-ORM-Llama3.1-70BModel Type=Open Reward Model, Backbone=Llama-3.1, Parameter Count=70B2025.07 | 76.5 | — | — | — | — | — | — | |
| llama-3.1-70B-Instruct-RM-RB2Backbone=Llama-3.1-70B-Instruct2025.07 | 76 | 81.3 | 41.9 | 69.9 | 88.4 | 86.5 | 88.3 | |
| Skywork-Reward-Gemma-2-27BBackbone=Gemma-2-27B2025.07 | 75.8 | 73.7 | 40.3 | 70.5 | 94.2 | 93.2 | 82.6 | |
| Skywork-Reward-V2-Qwen3-4BBackbone=Qwen3, Parameters=4B2025.07 | 75.5 | 77.3 | 46.2 | 73.2 | 92.2 | 96.6 | 67.4 | |
| Skywork-Reward-V2-Qwen3-4BModel Type=Reward Model, Backbone=Qwen3, Parameter Count=4B2025.07 | 75.5 | — | — | — | — | — | — | |
| claude-3-7-sonnet-20250219Release Date=202502192025.07 | 75.4 | 73.3 | 54.4 | 75 | 90.3 | 92.1 | 67.2 | |
| Skywork-Reward-Gemma-2-27B-v0.2Backbone=Gemma-2-27B, Version=v0.22025.07 | 75.3 | 76.7 | 37.5 | 67.2 | 96.9 | 91.7 | 81.8 | |
| Skywork-Reward-Gemma-2-27B-v0.2Model Type=Open Reward Model, Backbone=Gemma-2, Parameter Count=27B2025.07 | 75.3 | — | — | — | — | — | — | |
| Skywork-Reward-V2-Llama-3.2-3BBackbone=Llama-3.2, Parameters=3B2025.07 | 74.7 | 76.2 | 45.6 | 69.4 | 93.1 | 96 | 67.7 | |
| Skywork-Reward-V2-Llama-3.2-3BModel Type=Reward Model, Backbone=Llama-3.2, Parameter Count=3B2025.07 | 74.7 | — | — | — | — | — | — | |
| TCRMModel=Qwen3-32B2026.04 | 74.4 | 73.7 | 38.1 | 70.5 | 92.9 | 93.9 | 77.3 | |
| URM-LLama-3.1-8BBackbone=Llama-3.1-8B2025.07 | 73.9 | 68.8 | 45 | 63.9 | 91.8 | 97.6 | 76.5 | |
| BaselineModel=Qwen3-32B2026.04 | 73.4 | 72.6 | 38.1 | 68.9 | 93.1 | 91.1 | 76.4 | |
| Skywork-Reward-Llama-3.1-8BBackbone=Llama-3.1-8B2025.07 | 73.1 | 69.9 | 42.5 | 62.8 | 93.3 | 96.2 | 74.1 | |
| TC-λModel=Qwen3-14B2026.04 | 72.6 | 67.8 | 43.1 | 70.5 | 90.9 | 85.5 | 77.8 | |
| LDL-Reward-Gemma-2-27B-v0.1Model Type=Open Reward Model, Backbone=Gemma-2, Parameter Count=27B2025.07 | 72.5 | — | — | — | — | — | — | |
| BaselineModel=Qwen3-14B2026.04 | 72.2 | 67.6 | 42.5 | 70.5 | 90.2 | 80.8 | 81.5 | |
| Skywork-Reward-Llama-3.1-8B-v0.2Model Type=Open Reward Model, Backbone=Llama-3.1, Parameter Count=8B2025.07 | 71.8 | — | — | — | — | — | — | |
| TCRMModel=Qwen3-14B2026.04 | 71.1 | 66.5 | 41.2 | 65.6 | 93.6 | 85.1 | 74.5 | |
| Skywork-Reward-V2-Qwen3-1.7BBackbone=Qwen3, Parameters=1.7B2025.07 | 68.3 | 65.8 | 45 | 72.7 | 89.1 | 88.5 | 48.7 | |
| Skywork-Reward-V2-Qwen3-1.7BModel Type=Reward Model, Backbone=Qwen3, Parameter Count=1.7B2025.07 | 68.3 | — | — | — | — | — | — | |
| TCRMModel=Llama3.1-8B2026.04 | 68.1 | 60 | 33.1 | 67.2 | 87.8 | 87.1 | 73.3 | |
| TC-λModel=Llama3.1-8B2026.04 | 67.6 | 61.1 | 35 | 59 | 88.2 | 89.1 | 73.1 | |
| BaselineModel=Llama3.1-8B2026.04 | 67.3 | 59.8 | 33.8 | 65.6 | 88.4 | 87.5 | 69 | |
| ArmoRM-Llama3-8B-v0.1Model Type=Open Reward Model, Backbone=Llama-3, Parameter Count=8B2025.07 | 66.5 | — | — | — | — | — | — | |
| GPT-4oModel Type=LLM-as-a-Judge2025.07 | 64.9 | — | — | — | — | — | — | |
| Llama-3-OffsetBias-RM-8BModel Type=Open Reward Model, Backbone=Llama-3, Parameter Count=8B2025.07 | 64.8 | — | — | — | — | — | — | |
| Claude-3.5-SonnetModel Type=LLM-as-a-Judge2025.07 | 64.7 | — | — | — | — | — | — | |
| Skywork/Skywork-Reward-V2-Llama-3.2-1BParams (B)=1.0, Type=Seq. Classifier2026.02 | 64.376 | — | — | — | — | — | — | |
| Skywork-Reward-V2-Llama-3.2-1BBackbone=Llama-3.2, Parameters=1B2025.07 | 64.3 | 60.9 | 45.6 | 59.6 | 87.3 | 89.3 | 43.1 | |
| Skywork-Reward-V2-Llama-3.2-1BModel Type=Reward Model, Backbone=Llama-3.2, Parameter Count=1B2025.07 | 64.3 | — | — | — | — | — | — | |
| Skywork-Reward-V2-Qwen3-0.6BBackbone=Qwen3, Parameters=0.6B2025.07 | 61.3 | 58.2 | 40 | 71.6 | 84.4 | 79.4 | 34 | |
| Skywork-Reward-V2-Qwen3-0.6BModel Type=Reward Model, Backbone=Qwen3, Parameter Count=0.6B2025.07 | 61.3 | — | — | — | — | — | — | |
| Skywork/Skywork-Reward-V2-Qwen3-0.6BParams (B)=0.6, Type=Seq. Classifier2026.02 | 61.25 | — | — | — | — | — | — | |
| Internlm2-20b-rewardModel Type=Open Reward Model, Backbone=Internlm2, Parameter Count=20B2025.07 | 56.3 | — | — | — | — | — | — | |
| ImplicitPRM (DPO)Model=Qwen3-14B2026.04 | 54.1 | 53.9 | 31.2 | 55.7 | 76.2 | 83.8 | 23.9 | |
| TCRMModel=Qwen3-8B2026.04 | 53.2 | 48.2 | 37.5 | 57.4 | 74.4 | 77 | 24.5 | |
| ImplicitPRM (DPO)Model=Llama3.1-8B2026.04 | 52.5 | 50.9 | 23.8 | 54.1 | 82.7 | 86.1 | 17.5 | |
| ImplicitPRM (DPO)Model=Qwen3-32B2026.04 | 52 | 59.4 | 26.2 | 48.1 | 64.9 | 82.4 | 30.8 | |
| ImplicitPRM (DPO)Model=Qwen3-8B2026.04 | 51.7 | 49.1 | 31.9 | 54.6 | 66 | 82 | 26.4 | |
| ImplicitPRM (DPO)Model=Qwen3-4B2026.04 | 51 | 48.8 | 29.4 | 53 | 67.1 | 80.2 | 27.4 | |
| TC-λModel=Qwen3-8B2026.04 | 48.5 | 58.7 | 35 | 55.7 | 72.7 | 47.5 | 21.1 | |
| BaselineModel=Qwen3-8B2026.04 | 47.4 | 40.2 | 33.8 | 56.3 | 57.8 | 75.6 | 20.9 | |
| TC-λModel=Qwen3-4B2026.04 | 45.4 | 41.5 | 32.5 | 49.2 | 74 | 60.6 | 14.4 | |
| TC-λModel=Qwen3-1.7B2026.04 | 44.5 | 51 | 33.1 | 56.8 | 66.4 | 55 | 4.9 | |
| ImplicitPRM (DPO)Model=Qwen3-1.7B2026.04 | 42.1 | 46.5 | 21.2 | 50.8 | 52.4 | 69.1 | 12.5 | |
| TCRMModel=Qwen3-1.7B2026.04 | 37.2 | 33.7 | 27.5 | 29.5 | 62 | 64.4 | 5.9 | |
| ImplicitPRM (DPO)Model=Qwen3-0.6B2026.04 | 35.5 | 38.3 | 25 | 38.3 | 32.4 | 70.9 | 7.9 | |
| Finemath-RM-Llama-3.2-1BParams (B)=1.0, Type=Seq. Classifier2026.02 | 33.2 | — | — | — | — | — | — | |
| TC-λModel=Qwen3-0.6B2026.04 | 32.8 | 31.2 | 24.4 | 35 | 40.2 | 54.1 | 12 | |
| TCRMModel=Qwen3-0.6B2026.04 | 29 | 30.5 | 26.2 | 32.8 | 29.3 | 50.9 | 4.1 | |
| BaselineModel=Qwen3-1.7B2026.04 | 28.7 | 27.6 | 23.1 | 40.4 | 44.4 | 30.3 | 6.1 | |
| OpenAssistant/oasst-rm-2.1-pythia-1.4b-epoch-2.5Params (B)=1.4, Type=Seq. Classifier2026.02 | 26.479 | — | — | — | — | — | — | |
| BaselineModel=Qwen3-0.6B2026.04 | 22.4 | 37.5 | 28.1 | 25.1 | 21.6 | 16.4 | 5.5 |