← All comparisons

Claude Sonnet 4.6 (Non-reasoning, High Effort) vs Granite 3.3 8B (Non-reasoning)

Anthropic vs IBM — side-by-side benchmark comparison

Claude Sonnet 4.6 (Non-reasoning, High Effort)Granite 3.3 8B (Non-reasoning)
Intelligence Index44.47.0
Coding Index46.43.4
Math Index6.7
Output speed (tok/s)55.2453.9
Blended price ($/1M)$6.56$0.09
Time to first token (s)1.07s21.19s
aime4.7%
aime 256.7%
artificial analysis coding index46.403.40
artificial analysis intelligence index44.407.00
artificial analysis math index6.70
gpqa79.9%33.8%
hle13.2%4.2%
ifbench41.2%22.4%
lcr57.7%4.3%
livecodebench12.7%
math 50066.5%
mmlu pro46.8%
scicode46.9%10.1%
tau279.5%10.5%
terminalbench hard46.2%0.0%

Benchmark data from Artificial Analysis.