List your company
ResearchFormerly WizWand

General Instruction Following on Arena-Hard v2

85.9Scoreo3
-1.77220.98943.7566.511Jul 2, 2025Sep 1, 2025Nov 1, 2025Jan 2, 2026Mar 4, 2026May 4, 2026Jul 5, 2026
Updated 4mo ago

Evaluation results

MethodLinks
o3Type=Closed-source LLM2026.0285.9
Gemini-2.5-ProType=Closed-source LLM2026.0279
Qwen3-30B-A3B-Instruct + More Query Rubricsbackbone=Qwen3-30B-A3B-Instruct, supervision=More Query Rubrics2026.0274.6
Nemotron-3-Puzzle-75B-A9BModel=Nemotron-3-Puzzle-75B-A9B, Precision=FP82026.0769.8
GenRM-R-Align-14BReward Model=GenRM-R-Align-14B2026.0260.2
Claude-3.7-SonnetType=Closed-source LLM2026.0259.8
GenRM-R-Align-8BReward Model=GenRM-R-Align-8B2026.0259.5
Deepseek-R1Type=Open-source LLM2026.0256.8
GenRM-RLVR-14BReward Model=GenRM-RLVR-14B2026.0255.9
GenRM-RLVR-8BReward Model=GenRM-RLVR-8B2026.0253.1
Qwen3-14B-as-GenRMReward Model=Qwen3-14B-as-GenRM2026.0251
GPT-4.1Type=Closed-source LLM2026.0250
Qwen3-235B-InstructType=Open-source LLM2026.0246.7
Qwen3-8B-as-GenRMReward Model=Qwen3-8B-as-GenRM2026.0246.6
Baichuan-M2-32BType=Specialized LLM2026.0245.8
Qwen3-32BType=Open-source LLM2026.0244.5
HuatuoGPT-o1-72BType=Specialized LLM2026.0243.2
Qwen3-4B-Instruct + More Query Rubricsbackbone=Qwen3-4B-Instruct, supervision=More Query Rubrics2026.0241.2
Qwen3-4B-Instruct + Doctor Rubricsbackbone=Qwen3-4B-Instruct, supervision=Doctor Rubrics2026.0239.7
Qwen3-4B-Instruct + Principle Rubricsbackbone=Qwen3-4B-Instruct, supervision=Principle Rubrics2026.0237
Qwen3-4B-Instruct + Draft Rubricsbackbone=Qwen3-4B-Instruct, supervision=Draft Rubrics2026.0234.9
Qwen3-30B-A3B-InstructType=Our Method baseline2026.0233.9
Qwen3-8BReward Model=Baseline2026.0226.5
+RL (Skywork-Reward-V2-Llama-3.1-8B)Model=Qwen2.5-7B2025.0718.5
+RL (Skywork-Reward-V2-Qwen3-4B)Model=Qwen2.5-7B2025.0717.9
Instruct (official)Model=Qwen2.5-7B2025.0717.1
+RL (Skywork-Reward-Gemma-2-27B-v0.2)Model=Qwen2.5-7B2025.0715.5
Qwen3-4B-InstructType=Our Method baseline2026.0215
+RL (Skywork-Reward-Llama-3-8B-v0.2)Model=Qwen2.5-7B2025.0712.2
+SFTModel=Qwen2.5-7B2025.079.9
+RL (Skywork-Reward-V2-Llama-3.1-8B)Model=Llama-3.1-8B2025.076.3
+RL (Skywork-Reward-V2-Qwen3-4B)Model=Llama-3.1-8B2025.076
Instruct (official)Model=Llama-3.1-8B2025.075.8
BaseModel=Qwen2.5-7B2025.075.6
+RL (Skywork-Reward-Gemma-2-27B-v0.2)Model=Llama-3.1-8B2025.073.8
+SFTModel=Llama-3.1-8B2025.073.1
BaseModel=Llama-3.1-8B2025.072
+RL (Skywork-Reward-Llama-3-8B-v0.2)Model=Llama-3.1-8B2025.071.6