← All comparisons

Llama 3.1 Nemotron Ultra 253B v1 (Reasoning) vs DeepSeek V3.1 Terminus (Reasoning)

NVIDIA vs DeepSeek — side-by-side benchmark comparison

Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)DeepSeek V3.1 Terminus (Reasoning)
Intelligence Index15.033.9
Coding Index13.133.7
Math Index63.789.7
Output speed (tok/s)52.30.0
Blended price ($/1M)$0.90$1.91
Time to first token (s)0.76s0.00s
aime74.7%
aime 2563.7%89.7%
artificial analysis coding index13.1033.70
artificial analysis intelligence index15.0033.90
artificial analysis math index63.7089.70
gpqa72.8%79.2%
hle8.1%15.2%
ifbench38.2%57.0%
lcr7.3%65.0%
livecodebench64.1%79.8%
math 50095.2%
mmlu pro82.5%85.1%
scicode34.7%40.6%
tau211.4%37.1%
terminalbench hard2.3%30.3%

Benchmark data from Artificial Analysis.