← All comparisons

Llama 3.1 Nemotron Instruct 70B vs Granite 4.0 H 1B

NVIDIA vs IBM — side-by-side benchmark comparison

Llama 3.1 Nemotron Instruct 70BGranite 4.0 H 1B
Intelligence Index13.48.0
Coding Index10.82.7
Math Index11.06.3
Output speed (tok/s)292.90.0
Blended price ($/1M)$1.20$0.00
Time to first token (s)0.24s0.00s
aime24.7%
aime 2511.0%6.3%
artificial analysis coding index10.802.70
artificial analysis intelligence index13.408.00
artificial analysis math index11.006.30
gpqa46.5%26.3%
hle4.6%5.0%
ifbench30.7%26.2%
lcr7.0%6.3%
livecodebench16.9%11.5%
math 50073.3%
mmlu pro69.0%27.7%
scicode23.3%8.2%
tau223.1%19.6%
terminalbench hard4.5%0.0%

Benchmark data from Artificial Analysis.