← All comparisons

Llama 3.1 Nemotron Ultra 253B v1 (Reasoning) vs DeepSeek R1 Distill Qwen 32B

NVIDIA vs DeepSeek — side-by-side benchmark comparison

Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)DeepSeek R1 Distill Qwen 32B
Intelligence Index15.017.2
Coding Index13.1
Math Index63.763.0
Output speed (tok/s)52.30.0
Blended price ($/1M)$0.90$0.00
Time to first token (s)0.76s0.00s
aime74.7%68.7%
aime 2563.7%63.0%
artificial analysis coding index13.10
artificial analysis intelligence index15.0017.20
artificial analysis math index63.7063.00
gpqa72.8%61.5%
hle8.1%5.5%
ifbench38.2%22.9%
lcr7.3%9.7%
livecodebench64.1%27.0%
math 50095.2%94.1%
mmlu pro82.5%73.9%
scicode34.7%37.6%
tau211.4%
terminalbench hard2.3%

Benchmark data from Artificial Analysis.