← All comparisons

Hermes 4 - Llama-3.1 70B (Non-reasoning) vs DeepSeek R1 Distill Llama 70B

Nous Research vs DeepSeek — side-by-side benchmark comparison

Hermes 4 - Llama-3.1 70B (Non-reasoning)DeepSeek R1 Distill Llama 70B
Intelligence Index12.616.0
Coding Index9.211.4
Math Index11.353.7
Output speed (tok/s)94.346.8
Blended price ($/1M)$0.20$0.79
Time to first token (s)0.61s0.33s
aime67.0%
aime 2511.3%53.7%
artificial analysis coding index9.2011.40
artificial analysis intelligence index12.6016.00
artificial analysis math index11.3053.70
gpqa49.1%40.2%
hle3.6%6.1%
ifbench29.0%27.6%
lcr2.0%11.0%
livecodebench26.9%26.6%
math 50093.5%
mmlu pro66.4%79.5%
scicode27.7%31.3%
tau221.6%21.9%
terminalbench hard0.0%1.5%

Benchmark data from Artificial Analysis.