← All comparisons

DeepSeek V4 Flash (Non-reasoning) vs Claude 3.5 Sonnet (Oct '24)

DeepSeek vs Anthropic — side-by-side benchmark comparison

DeepSeek V4 Flash (Non-reasoning)Claude 3.5 Sonnet (Oct '24)
Intelligence Index36.515.9
Coding Index35.230.2
Math Index
Output speed (tok/s)127.90.0
Blended price ($/1M)$0.17$6.56
Time to first token (s)0.90s0.00s
aime15.7%
aime 25
artificial analysis coding index35.2030.20
artificial analysis intelligence index36.5015.90
artificial analysis math index
gpqa71.6%59.9%
hle7.0%3.9%
ifbench47.2%
lcr33.3%
livecodebench38.1%
math 50077.1%
mmlu pro77.2%
scicode37.3%36.6%
tau294.4%
terminalbench hard34.1%

Benchmark data from Artificial Analysis.