← All comparisons

Hermes 4 - Llama-3.1 405B (Reasoning) vs DeepSeek V3.2 Speciale

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

Hermes 4 - Llama-3.1 405B (Reasoning)DeepSeek V3.2 Speciale
Intelligence Index18.629.4
Coding Index16.037.9
Math Index69.796.7
Output speed (tok/s)38.60.0
Blended price ($/1M)$1.50$0.00
Time to first token (s)0.79s0.00s
aime
aime 2569.7%96.7%
artificial analysis coding index16.0037.90
artificial analysis intelligence index18.6029.40
artificial analysis math index69.7096.70
gpqa72.7%87.1%
hle10.3%26.1%
ifbench32.7%63.9%
lcr20.7%59.3%
livecodebench68.6%89.6%
math 500
mmlu pro82.9%86.3%
scicode25.2%44.0%
tau222.2%0.0%
terminalbench hard11.4%34.8%

Benchmark data from Artificial Analysis.