# MMLU leaderboard

Knowledge benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelscale.dev/leaderboards/benchmarks/knowledge-mmlu

Massive Multitask Language Understanding. Every model the catalog carries a published MMLU value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Knowledge |
| Measure | Multiple choice questions |
| Tasks | 57 subjects |
| Difficulty | Elementary to professional level |
| Published by | [Measuring Massive Multitask Language Understanding](https://arxiv.org/abs/2009.03300) |

A comprehensive multiple-choice question answering test covering 57 tasks including elementary mathematics, US history, computer science, law, and more. Tests knowledge across diverse academic subjects from high school to professional level.

## MMLU ranking

35 models with a published MMLU value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Multiple choice questions | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [o1](https://modelscale.dev/models/o1) | OpenAI | 91.8 | Observed 2026-11-21 · source benchlm:models |
| 2 | [DeepSeek-R1](https://modelscale.dev/models/deepseek-r1) | DeepSeek | 90.8 | Observed 2026-11-21 · source benchlm:models |
| 2 | [GPT-4.5](https://modelscale.dev/models/gpt-4-5) | OpenAI | 90.8 | Observed 2026-11-21 · source benchlm:models |
| 4 | [Sarvam 105B](https://modelscale.dev/models/sarvam-105b) | Sarvam | 90.6 | Observed 2026-11-21 · source benchlm:models |
| 5 | [GPT-4.1](https://modelscale.dev/models/gpt-4-1) | OpenAI | 90.2 | Observed 2026-11-21 · source benchlm:models |
| 6 | [GPT-OSS 120B](https://modelscale.dev/models/gpt-oss-120b) | OpenAI | 90 | Observed 2026-11-21 · source benchlm:models |
| 6 | [o4-mini](https://modelscale.dev/models/o4-mini) | OpenAI | 90 | Observed 2026-11-21 · source benchlm:models |
| 8 | [Ripple-65B Preview](https://modelscale.dev/models/ripple-65b-preview) | Ricursive Intelligence | 89.6 | Observed 2026-11-21 · source benchlm:models |
| 9 | [Kimi K2](https://modelscale.dev/models/kimi-k2) | Moonshot AI | 89.5 | Observed 2026-11-21 · source benchlm:models |
| 10 | [MiniMax-Text-01](https://modelscale.dev/models/minimax-text-01) | MiniMax | 88.5 | Observed 2026-11-21 · source benchlm:models |
| 11 | [GPT-4.1 mini](https://modelscale.dev/models/gpt-4-1-mini) | OpenAI | 87.5 | Observed 2026-11-21 · source benchlm:models |
| 11 | [Grok 2](https://modelscale.dev/models/grok-2) | xAI | 87.5 | Observed 2026-11-21 · source benchlm:models |
| 13 | [Trinity-Large-Preview](https://modelscale.dev/models/trinity-large-preview) | Arcee AI | 87.2 | Observed 2026-11-21 · source benchlm:models |
| 14 | [o3-mini](https://modelscale.dev/models/o3-mini) | OpenAI | 86.9 | Observed 2026-11-21 · source benchlm:models |
| 15 | [Grok 2 mini](https://modelscale.dev/models/grok-2-mini) | xAI | 86.2 | Observed 2026-11-21 · source benchlm:models |
| 16 | [Particle-1](https://modelscale.dev/models/particle-1) | Particle AI | 86.08 | Observed 2026-11-21 · source benchlm:models |
| 17 | [Llama 3.1 70B](https://modelscale.dev/models/llama-3-1-70b) | Meta | 86 | Observed 2026-11-21 · source benchlm:models |
| 17 | [Llama 3.3 70B](https://modelscale.dev/models/llama-3-3-70b) | Meta | 86 | Observed 2026-11-21 · source benchlm:models |
| 19 | [LongCat-Flash-Lite-Sparse](https://modelscale.dev/models/longcat-flash-lite-sparse) | Meituan | 85.31 | Observed 2026-11-21 · source benchlm:models |
| 20 | [GPT-OSS 20B](https://modelscale.dev/models/gpt-oss-20b) | OpenAI | 85.3 | Observed 2026-11-21 · source benchlm:models |
| 21 | [o1-mini](https://modelscale.dev/models/o1-mini) | OpenAI | 85.2 | Observed 2026-11-21 · source benchlm:models |
| 22 | [Sarvam 30B](https://modelscale.dev/models/sarvam-30b) | Sarvam | 85.1 | Observed 2026-11-21 · source benchlm:models |
| 23 | [Phi-4](https://modelscale.dev/models/phi-4) | Microsoft | 84.8 | Observed 2026-11-21 · source benchlm:models |
| 24 | [GPT-4o mini](https://modelscale.dev/models/gpt-4o-mini) | OpenAI | 82 | Observed 2026-11-21 · source benchlm:models |
| 25 | [Mistral Small 3.1](https://modelscale.dev/models/mistral-small-3-1) | Mistral | 80.62 | Observed 2026-11-21 · source benchlm:models |
| 26 | [Claude 3.5 Haiku](https://modelscale.dev/models/claude-3-5-haiku) | Anthropic | 80.3 | Observed 2026-11-21 · source benchlm:models |
| 27 | [GPT-4.1 nano](https://modelscale.dev/models/gpt-4-1-nano) | OpenAI | 80.1 | Observed 2026-11-21 · source benchlm:models |
| 28 | [Mixtral 8x22B Instruct v0.1](https://modelscale.dev/models/mixtral-8x22b-instruct-v0-1) | Mistral | 77.8 | Observed 2026-11-21 · source benchlm:models |
| 29 | [Gemma 2 27B](https://modelscale.dev/models/gemma-2-27b) | Google | 76.2 | Observed 2026-11-21 · source benchlm:models |
| 30 | [Llama 3.1 8B](https://modelscale.dev/models/llama-3-1-8b) | Meta | 73 | Observed 2026-11-21 · source benchlm:models |
| 31 | [Phi-3.5-mini](https://modelscale.dev/models/phi-3-5-mini) | Microsoft | 69 | Observed 2026-11-21 · source benchlm:models |
| 32 | [Granite-4.0-H-1B](https://modelscale.dev/models/granite-4-0-h-1b) | IBM | 59.74 | Observed 2026-11-21 · source benchlm:models |
| 33 | [Granite-4.0-1B](https://modelscale.dev/models/granite-4-0-1b) | IBM | 59.39 | Observed 2026-11-21 · source benchlm:models |
| 34 | [Granite-4.0-H-350M](https://modelscale.dev/models/granite-4-0-h-350m) | IBM | 36.21 | Observed 2026-11-21 · source benchlm:models |
| 35 | [Granite-4.0-350M](https://modelscale.dev/models/granite-4-0-350m) | IBM | 35.01 | Observed 2026-11-21 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
