← All comparisons

Claude 4.5 Haiku (Reasoning) vs DeepSeek V3.1 (Reasoning)

Anthropic vs DeepSeek — side-by-side benchmark comparison

Claude 4.5 Haiku (Reasoning)DeepSeek V3.1 (Reasoning)
Intelligence Index37.127.7
Coding Index32.629.7
Math Index83.789.7
Output speed (tok/s)142.20.0
Blended price ($/1M)$2.19$0.86
Time to first token (s)10.48s0.00s
aime
aime 2583.7%89.7%
artificial analysis coding index32.6029.70
artificial analysis intelligence index37.1027.70
artificial analysis math index83.7089.70
gpqa67.2%77.9%
hle9.7%13.0%
ifbench54.3%41.5%
lcr70.3%53.3%
livecodebench61.5%78.4%
math 500
mmlu pro76.0%85.1%
scicode43.3%39.1%
tau254.7%37.4%
terminalbench hard27.3%25.0%

Benchmark data from Artificial Analysis.