← All comparisons

Claude 4.5 Haiku (Reasoning) vs Granite 3.3 8B (Non-reasoning)

Anthropic vs IBM — side-by-side benchmark comparison

Claude 4.5 Haiku (Reasoning)Granite 3.3 8B (Non-reasoning)
Intelligence Index37.17.0
Coding Index32.63.4
Math Index83.76.7
Output speed (tok/s)142.2453.9
Blended price ($/1M)$2.19$0.09
Time to first token (s)10.48s21.19s
aime4.7%
aime 2583.7%6.7%
artificial analysis coding index32.603.40
artificial analysis intelligence index37.107.00
artificial analysis math index83.706.70
gpqa67.2%33.8%
hle9.7%4.2%
ifbench54.3%22.4%
lcr70.3%4.3%
livecodebench61.5%12.7%
math 50066.5%
mmlu pro76.0%46.8%
scicode43.3%10.1%
tau254.7%10.5%
terminalbench hard27.3%0.0%

Benchmark data from Artificial Analysis.