# General Instruction Following on Arena-Hard v2 Source: [General Instruction Following on Arena-Hard v2]() Collection index: [llms.txt]() Data updated: 2026-07-07T10:25:37.921Z Evaluation of model performance on general instruction following tasks using the Arena-Hard v2 dataset, measured by Score. 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 | General Instruction Following | | Dataset (including any stated split) | Arena-Hard v2 | | Metric | Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 38 of 38 | | 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 - [Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Score | Result date | Report paper | | --- | --- | --- | --- | | o3 | 85.9 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Gemini-2.5-Pro | 79 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-30B-A3B-Instruct + More Query Rubrics | 74.6 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Nemotron-3-Puzzle-75B-A9B | 69.8 | Not recorded | 6a4c882e85dba69fef8b2b88 | | GenRM-R-Align-14B | 60.2 | Not recorded | 698969fcc97a0b3e3b5dc4bb | | Claude-3.7-Sonnet | 59.8 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | GenRM-R-Align-8B | 59.5 | Not recorded | 698969fcc97a0b3e3b5dc4bb | | Deepseek-R1 | 56.8 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | GenRM-RLVR-14B | 55.9 | Not recorded | 698969fcc97a0b3e3b5dc4bb | | GenRM-RLVR-8B | 53.1 | Not recorded | 698969fcc97a0b3e3b5dc4bb | | Qwen3-14B-as-GenRM | 51 | Not recorded | 698969fcc97a0b3e3b5dc4bb | | GPT-4.1 | 50 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-235B-Instruct | 46.7 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-8B-as-GenRM | 46.6 | Not recorded | 698969fcc97a0b3e3b5dc4bb | | Baichuan-M2-32B | 45.8 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-32B | 44.5 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | HuatuoGPT-o1-72B | 43.2 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-4B-Instruct + More Query Rubrics | 41.2 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-4B-Instruct + Doctor Rubrics | 39.7 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-4B-Instruct + Principle Rubrics | 37 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-4B-Instruct + Draft Rubrics | 34.9 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-30B-A3B-Instruct | 33.9 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | Qwen3-8B | 26.5 | Not recorded | 698969fcc97a0b3e3b5dc4bb | | +RL (Skywork-Reward-V2-Llama-3.1-8B) | 18.5 | Not recorded | 68fa79c61e209131cea7f4fe | | +RL (Skywork-Reward-V2-Qwen3-4B) | 17.9 | Not recorded | 68fa79c61e209131cea7f4fe | | Instruct (official) | 17.1 | Not recorded | 68fa79c61e209131cea7f4fe | | +RL (Skywork-Reward-Gemma-2-27B-v0.2) | 15.5 | Not recorded | 68fa79c61e209131cea7f4fe | | Qwen3-4B-Instruct | 15 | Not recorded | 698c0d57c97a0b3e3b5e13a3 | | +RL (Skywork-Reward-Llama-3-8B-v0.2) | 12.2 | Not recorded | 68fa79c61e209131cea7f4fe | | +SFT | 9.9 | Not recorded | 68fa79c61e209131cea7f4fe | | +RL (Skywork-Reward-V2-Llama-3.1-8B) | 6.3 | Not recorded | 68fa79c61e209131cea7f4fe | | +RL (Skywork-Reward-V2-Qwen3-4B) | 6 | Not recorded | 68fa79c61e209131cea7f4fe | | Instruct (official) | 5.8 | Not recorded | 68fa79c61e209131cea7f4fe | | Base | 5.6 | Not recorded | 68fa79c61e209131cea7f4fe | | +RL (Skywork-Reward-Gemma-2-27B-v0.2) | 3.8 | Not recorded | 68fa79c61e209131cea7f4fe | | +SFT | 3.1 | Not recorded | 68fa79c61e209131cea7f4fe | | Base | 2 | Not recorded | 68fa79c61e209131cea7f4fe | | +RL (Skywork-Reward-Llama-3-8B-v0.2) | 1.6 | Not recorded | 68fa79c61e209131cea7f4fe | ### o3 Score: 85.9. Result ID: 6995f6c85698665ec42ab25c. [Reporting paper]() Recorded evaluation configuration: {"Type":"Closed-source LLM"}. This record does not assert independent reproduction. ### Gemini-2.5-Pro Score: 79. Result ID: 6995f6c85698665ec42ab25e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"Closed-source LLM"}. This record does not assert independent reproduction. ### Qwen3-30B-A3B-Instruct + More Query Rubrics Score: 74.6. Result ID: 6995f6c85698665ec42ab268. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"backbone":"Qwen3-30B-A3B-Instruct","supervision":"More Query Rubrics"}. This record does not assert independent reproduction. ### Nemotron-3-Puzzle-75B-A9B Score: 69.8. Result ID: 6a4cd40c60b88bb339ca8902. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Nemotron-3-Puzzle-75B-A9B","Precision":"FP8"}. This record does not assert independent reproduction. ### GenRM-R-Align-14B Score: 60.2. Result ID: 699644525698665ec488037e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Reward Model":"GenRM-R-Align-14B"}. This record does not assert independent reproduction. ### Claude-3.7-Sonnet Score: 59.8. Result ID: 6995f6c85698665ec42ab25d. [Reporting paper]() Recorded evaluation configuration: {"Type":"Closed-source LLM"}. This record does not assert independent reproduction. ### GenRM-R-Align-8B Score: 59.5. Result ID: 699644525698665ec488037b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Reward Model":"GenRM-R-Align-8B"}. This record does not assert independent reproduction. ### Deepseek-R1 Score: 56.8. Result ID: 6995f6c85698665ec42ab259. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"Open-source LLM"}. This record does not assert independent reproduction. ### GenRM-RLVR-14B Score: 55.9. Result ID: 699644525698665ec488037d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Reward Model":"GenRM-RLVR-14B"}. This record does not assert independent reproduction. ### GenRM-RLVR-8B Score: 53.1. Result ID: 699644525698665ec488037a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Reward Model":"GenRM-RLVR-8B"}. This record does not assert independent reproduction. ### Qwen3-14B-as-GenRM Score: 51. Result ID: 699644525698665ec488037c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Reward Model":"Qwen3-14B-as-GenRM"}. This record does not assert independent reproduction. ### GPT-4.1 Score: 50. Result ID: 6995f6c85698665ec42ab25f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Type":"Closed-source LLM"}. This record does not assert independent reproduction. ### Qwen3-235B-Instruct Score: 46.7. Result ID: 6995f6c85698665ec42ab25a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"Open-source LLM"}. This record does not assert independent reproduction. ### Qwen3-8B-as-GenRM Score: 46.6. Result ID: 699644525698665ec4880379. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Reward Model":"Qwen3-8B-as-GenRM"}. This record does not assert independent reproduction. ### Baichuan-M2-32B Score: 45.8. Result ID: 6995f6c85698665ec42ab260. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"Specialized LLM"}. This record does not assert independent reproduction. ### Qwen3-32B Score: 44.5. Result ID: 6995f6c85698665ec42ab25b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"Open-source LLM"}. This record does not assert independent reproduction. ### HuatuoGPT-o1-72B Score: 43.2. Result ID: 6995f6c85698665ec42ab261. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"Specialized LLM"}. This record does not assert independent reproduction. ### Qwen3-4B-Instruct + More Query Rubrics Score: 41.2. Result ID: 6995f6c85698665ec42ab266. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"backbone":"Qwen3-4B-Instruct","supervision":"More Query Rubrics"}. This record does not assert independent reproduction. ### Qwen3-4B-Instruct + Doctor Rubrics Score: 39.7. Result ID: 6995f6c85698665ec42ab264. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"backbone":"Qwen3-4B-Instruct","supervision":"Doctor Rubrics"}. This record does not assert independent reproduction. ### Qwen3-4B-Instruct + Principle Rubrics Score: 37. Result ID: 6995f6c85698665ec42ab265. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"backbone":"Qwen3-4B-Instruct","supervision":"Principle Rubrics"}. This record does not assert independent reproduction. ### Qwen3-4B-Instruct + Draft Rubrics Score: 34.9. Result ID: 6995f6c85698665ec42ab263. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"backbone":"Qwen3-4B-Instruct","supervision":"Draft Rubrics"}. This record does not assert independent reproduction. ### Qwen3-30B-A3B-Instruct Score: 33.9. Result ID: 6995f6c85698665ec42ab267. [Reporting paper]() Recorded evaluation configuration: {"Type":"Our Method baseline"}. This record does not assert independent reproduction. ### Qwen3-8B Score: 26.5. Result ID: 699644525698665ec4880378. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Reward Model":"Baseline"}. This record does not assert independent reproduction. ### +RL (Skywork-Reward-V2-Llama-3.1-8B) Score: 18.5. Result ID: 69a84d686dbb7075b94f6419. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### +RL (Skywork-Reward-V2-Qwen3-4B) Score: 17.9. Result ID: 69a84d686dbb7075b94f6418. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### Instruct (official) Score: 17.1. Result ID: 69a84d686dbb7075b94f641a. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### +RL (Skywork-Reward-Gemma-2-27B-v0.2) Score: 15.5. Result ID: 69a84d686dbb7075b94f6417. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### Qwen3-4B-Instruct Score: 15. Result ID: 6995f6c85698665ec42ab262. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Type":"Our Method baseline"}. This record does not assert independent reproduction. ### +RL (Skywork-Reward-Llama-3-8B-v0.2) Score: 12.2. Result ID: 69a84d686dbb7075b94f6416. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### +SFT Score: 9.9. Result ID: 69a84d686dbb7075b94f6415. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### +RL (Skywork-Reward-V2-Llama-3.1-8B) Score: 6.3. Result ID: 69a84d686dbb7075b94f6412. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B"}. This record does not assert independent reproduction. ### +RL (Skywork-Reward-V2-Qwen3-4B) Score: 6. Result ID: 69a84d686dbb7075b94f6411. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B"}. This record does not assert independent reproduction. ### Instruct (official) Score: 5.8. Result ID: 69a84d686dbb7075b94f6413. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B"}. This record does not assert independent reproduction. ### Base Score: 5.6. Result ID: 69a84d686dbb7075b94f6414. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### +RL (Skywork-Reward-Gemma-2-27B-v0.2) Score: 3.8. Result ID: 69a84d686dbb7075b94f6410. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B"}. This record does not assert independent reproduction. ### +SFT Score: 3.1. Result ID: 69a84d686dbb7075b94f640e. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B"}. This record does not assert independent reproduction. ### Base Score: 2. Result ID: 69a84d686dbb7075b94f640d. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B"}. This record does not assert independent reproduction. ### +RL (Skywork-Reward-Llama-3-8B-v0.2) Score: 1.6. Result ID: 69a84d686dbb7075b94f640f. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama-3.1-8B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.