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. | 671B | 49B |
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 | 128k |
| Benchmarks | ||
LiveCodeBenchCompetitive coding — Coding problems published so recently the AI can't have seen them in training — a contamination-free test of raw programming skill. Higher is better. | 88.7% | 73.58% |
Humanity's Last Exam · text only, no toolsMultidisciplinary reasoning — accuracy on the 2,158 text-only questions without tools | 30.6% | 7.64% |
AIME 2025Reasoning — single-answer accuracy on the 2025 invitational mathematics exam without tools | 96% | 82.71% |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 85.7% | 71.97% |
MATH-500Reasoning — accuracy on the 500-problem MATH subset | — | 97.4% |
AIME 2024Reasoning — single-answer accuracy on the 2024 invitational mathematics exam without tools | — | 87.5% |
MMLU-ProKnowledge — accuracy on the ten-option MMLU-Pro benchmark | — | 79.53% |
| Overview | ||
| Company | DeepSeek | NVIDIA |
| Release date | Dec 1 2025 | Jul 25 2025 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
DeepSeek-V3.2-SpecialevsClaude Fable 5.2Llama Nemotron Super 49B v1.5vsClaude Fable 5.2DeepSeek-V3.2-SpecialevsGPT-6.2 AstraLlama Nemotron Super 49B v1.5vsGPT-6.2 AstraDeepSeek-V3.2-SpecialevsGemini 4 NeonLlama Nemotron Super 49B v1.5vsGemini 4 NeonDeepSeek-V3.2-SpecialevsMuse EmberLlama Nemotron Super 49B v1.5vsMuse EmberDeepSeek-V3.2-SpecialevsGrok 4.7Llama Nemotron Super 49B v1.5vsGrok 4.7DeepSeek-V3.2-SpecialevsMistral Large 4Llama Nemotron Super 49B v1.5vsMistral Large 4Frequently asked questions
DeepSeek-V3.2-Speciale leads Llama Nemotron Super 49B v1.5 on 4 of the 4 benchmarks they both report (LiveCodeBench, Humanity's Last Exam, AIME 2025, GPQA Diamond). Llama Nemotron Super 49B v1.5 shipped 129 days before DeepSeek-V3.2-Speciale, so benchmark comparisons should account for the intervening progress.
DeepSeek-V3.2-Speciale has 671B parameters, while Llama Nemotron Super 49B v1.5 has 49B. Context windows are 128k (DeepSeek-V3.2-Speciale) vs 128k (Llama Nemotron Super 49B v1.5).
On LiveCodeBench, DeepSeek-V3.2-Speciale leads at 88.7% vs Llama Nemotron Super 49B v1.5 at 73.58%. On Humanity's Last Exam · text only, no tools, DeepSeek-V3.2-Speciale leads at 30.6% vs Llama Nemotron Super 49B v1.5 at 7.64%. On AIME 2025, DeepSeek-V3.2-Speciale leads at 96% vs Llama Nemotron Super 49B v1.5 at 82.71%. On GPQA Diamond, DeepSeek-V3.2-Speciale leads at 85.7% vs Llama Nemotron Super 49B v1.5 at 71.97%.