# Language Understanding on MMLU-Redux Generative Source: [Language Understanding on MMLU-Redux Generative]() Collection index: [llms.txt]() Data updated: 2026-05-11T10:11:09.984Z Evaluation of language understanding capabilities on the MMLU-Redux Generative dataset, reporting performance across specific subject areas like humanities, social sciences, and STEM. 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 Understanding | | Dataset (including any stated split) | MMLU-Redux Generative | | Metric | Humanities Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | 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 - [Humanities Accuracy](): Higher is better; unit: PERCENT. - [Other Accuracy](): Higher is better; unit: PERCENT. - [Social Sciences Accuracy](): Higher is better; unit: PERCENT. - [STEM Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Humanities Accuracy | Result date | Report paper | | --- | --- | --- | --- | | GPT-5 nano | 87.8 | Not recorded | 6a0162badbc836b87d82332a | | gpt-oss-20b | 85.8 | Not recorded | 6a0162badbc836b87d82332a | | Qwen3-8B | 84.3 | Not recorded | 6a0162badbc836b87d82332a | | Qwen3-4B | 81 | Not recorded | 6a0162badbc836b87d82332a | | Ministral-3-8B | 78.8 | Not recorded | 6a0162badbc836b87d82332a | | gemma-3-12b-it | 77.4 | Not recorded | 6a0162badbc836b87d82332a | | EngGPT2-16B-A3B | 72.5 | Not recorded | 6a0162badbc836b87d82332a | | Llama-3.1-8B-Instruct | 72.2 | Not recorded | 6a0162badbc836b87d82332a | | LLaMAntino-3 | 71.9 | Not recorded | 6a0162badbc836b87d82332a | | Moonlight-16B-A3B-Instruct | 71.5 | Not recorded | 6a0162badbc836b87d82332a | | gemma-3-4b-it | 64.3 | Not recorded | 6a0162badbc836b87d82332a | | Velvet-14B | 60.4 | Not recorded | 6a0162badbc836b87d82332a | | Llama-3.2-3B-Instruct | 59.5 | Not recorded | 6a0162badbc836b87d82332a | | FastwebMIIA-7B | 46.4 | Not recorded | 6a0162badbc836b87d82332a | | deepseek-moe-16b-chat | 40.2 | Not recorded | 6a0162badbc836b87d82332a | | Minerva-7B-instruct-v1.0 | 30.6 | Not recorded | 6a0162badbc836b87d82332a | ### GPT-5 nano Humanities Accuracy: 87.8. Result ID: 6a01ab3c47f34ce924e41f61. [Reporting paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### gpt-oss-20b Humanities Accuracy: 85.8. Result ID: 6a01ab3c47f34ce924e41f63. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### Qwen3-8B Humanities Accuracy: 84.3. Result ID: 6a01ab3c47f34ce924e41f5b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### Qwen3-4B Humanities Accuracy: 81. Result ID: 6a01ab3c47f34ce924e41f5a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### Ministral-3-8B Humanities Accuracy: 78.8. Result ID: 6a01ab3c47f34ce924e41f60. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### gemma-3-12b-it Humanities Accuracy: 77.4. Result ID: 6a01ab3c47f34ce924e41f5f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### EngGPT2-16B-A3B Humanities Accuracy: 72.5. Result ID: 6a01ab3c47f34ce924e41f55. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### Llama-3.1-8B-Instruct Humanities Accuracy: 72.2. Result ID: 6a01ab3c47f34ce924e41f5d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### LLaMAntino-3 Humanities Accuracy: 71.9. Result ID: 6a01ab3c47f34ce924e41f59. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### Moonlight-16B-A3B-Instruct Humanities Accuracy: 71.5. Result ID: 6a01ab3c47f34ce924e41f64. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### gemma-3-4b-it Humanities Accuracy: 64.3. Result ID: 6a01ab3c47f34ce924e41f5e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### Velvet-14B Humanities Accuracy: 60.4. Result ID: 6a01ab3c47f34ce924e41f58. [Reporting paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### Llama-3.2-3B-Instruct Humanities Accuracy: 59.5. Result ID: 6a01ab3c47f34ce924e41f5c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### FastwebMIIA-7B Humanities Accuracy: 46.4. Result ID: 6a01ab3c47f34ce924e41f56. [Reporting paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### deepseek-moe-16b-chat Humanities Accuracy: 40.2. Result ID: 6a01ab3c47f34ce924e41f62. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. ### Minerva-7B-instruct-v1.0 Humanities Accuracy: 30.6. Result ID: 6a01ab3c47f34ce924e41f57. [Reporting paper]() Recorded evaluation configuration: {"evaluation protocol":"generative","match type":"exact match"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.