← All comparisons

DeepSeek V3.1 (Non-reasoning) vs Llama 3.1 Tulu3 405B

DeepSeek vs Allen Institute for AI — side-by-side benchmark comparison

DeepSeek V3.1 (Non-reasoning)Llama 3.1 Tulu3 405B
Intelligence Index28.114.1
Coding Index28.4—
Math Index49.7—
Output speed (tok/s)0.00.0
Blended price ($/1M)$0.83$0.00
Time to first token (s)0.00s0.00s
aime—13.3%
aime 2549.7%—
artificial analysis coding index28.40—
artificial analysis intelligence index28.1014.10
artificial analysis math index49.70—
gpqa73.5%51.6%
hle6.3%3.5%
ifbench37.8%—
lcr45.0%—
livecodebench57.7%29.1%
math 500—77.8%
mmlu pro83.3%71.6%
scicode36.7%30.2%
tau234.8%—
terminalbench hard24.2%—

Benchmark data from Artificial Analysis.