Compare AI models
| Specifications | ||
ParametersThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. | 51B | — |
Context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider. | 128k | 200k |
| Benchmarks | ||
MMLUGeneral knowledge — A 57-subject multiple-choice exam — history, law, medicine, maths — that was the standard measure of how much a model knows from 2020 until roughly 2024, when frontier scores crowded into the high 80s and labs moved on to harder tests. The scores here were published years apart under different testing setups, so read them as a historical record rather than a like-for-like ranking. Higher is better. | 80.2% | 90.3% |
HumanEvalFunction synthesis — 164 small Python problems: the AI is given a function's description and has to write the working function. This was the coding benchmark of the GPT-3.5 and GPT-4 era, before the field moved to fixing real bugs in real repositories. Higher is better. | — | 99.3% |
MATH-500Reasoning — accuracy on the 500-problem MATH subset | — | 98.2% |
DROPReasoning — three-shot token F1 on questions requiring reasoning over paragraphs | — | 78.1 |
ARC-ChallengeReasoning — multiple-choice science questions from the AI2 Reasoning Challenge; unrelated to ARC-AGI | 69.2% | — |
HellaSwagReasoning — commonsense completion of short scenarios | 85.58% | — |
WinoGrandeReasoning — resolving ambiguous pronouns in context | 84.53% | — |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | — | 81.3% |
SimpleQAKnowledge — accuracy on short, fact-seeking questions in the original SimpleQA set | — | 19.3% |
MGSMMultilingual math — solving grade-school word problems translated into multiple languages | — | 93.5% |
GSM8KGrade-school math — Grade-school maths word problems that take a few steps of arithmetic to work through. It separated the models of 2022 and 2023 sharply, then saturated. One caveat on the historical numbers: OpenAI included part of the GSM8K training set in GPT-4's pre-training mix, so GPT-4's score is not a clean few-shot result. Higher is better. | 91.43% | — |
| Overview | ||
| Company | NVIDIA | OpenAI |
| Release date | Sep 23 2024 | Apr 16 2025 |
| Access | Open Weight | Closed |
| Model details | View model | View model |
Other comparisons
Llama 3.1 Nemotron 51BvsClaude Fable 5.2o4-mini-highvsClaude Fable 5.2Llama 3.1 Nemotron 51BvsGemini 4 Neono4-mini-highvsGemini 4 NeonLlama 3.1 Nemotron 51BvsMuse Embero4-mini-highvsMuse EmberLlama 3.1 Nemotron 51BvsGrok 4.7o4-mini-highvsGrok 4.7Llama 3.1 Nemotron 51BvsDeepSeek-V4.1-Flash-1017o4-mini-highvsDeepSeek-V4.1-Flash-1017Llama 3.1 Nemotron 51BvsMistral Large 4o4-mini-highvsMistral Large 4Frequently asked questions
o4-mini-high leads Llama 3.1 Nemotron 51B on 1 of the 1 benchmark they both report (MMLU). Llama 3.1 Nemotron 51B shipped 205 days before o4-mini-high, so benchmark comparisons should account for the intervening progress.
Context windows are 128k (Llama 3.1 Nemotron 51B) vs 200k (o4-mini-high). Llama 3.1 Nemotron 51B is open weight, while o4-mini-high is closed.
On MMLU, o4-mini-high leads at 90.3% vs Llama 3.1 Nemotron 51B at 80.2%.