← All comparisons

Claude 4.5 Sonnet (Reasoning) vs DeepSeek V3.2 (Reasoning)

Anthropic vs DeepSeek — side-by-side benchmark comparison

Claude 4.5 Sonnet (Reasoning)DeepSeek V3.2 (Reasoning)
Intelligence Index43.041.7
Coding Index38.636.7
Math Index88.092.0
Output speed (tok/s)55.00.0
Blended price ($/1M)$6.56$0.34
Time to first token (s)7.02s0.00s
aime
aime 2588.0%92.0%
artificial analysis coding index38.6036.70
artificial analysis intelligence index43.0041.70
artificial analysis math index88.0092.00
gpqa83.4%84.0%
hle17.3%22.2%
ifbench57.3%60.7%
lcr65.7%65.0%
livecodebench71.4%86.2%
math 500
mmlu pro87.5%86.2%
scicode44.7%38.9%
tau278.1%90.6%
terminalbench hard35.6%35.6%

Benchmark data from Artificial Analysis.