← All comparisons

Claude 4.5 Haiku (Reasoning) vs Claude Opus 4.6 (Adaptive Reasoning, Max Effort)

Anthropic vs Anthropic — side-by-side benchmark comparison

Claude 4.5 Haiku (Reasoning)Claude Opus 4.6 (Adaptive Reasoning, Max Effort)
Intelligence Index37.152.9
Coding Index32.648.1
Math Index83.7
Output speed (tok/s)142.254.8
Blended price ($/1M)$2.19$10.94
Time to first token (s)10.48s11.69s
aime
aime 2583.7%
artificial analysis coding index32.6048.10
artificial analysis intelligence index37.1052.90
artificial analysis math index83.70
gpqa67.2%89.6%
hle9.7%36.7%
ifbench54.3%53.1%
lcr70.3%70.7%
livecodebench61.5%
math 500
mmlu pro76.0%
scicode43.3%51.9%
tau254.7%92.1%
terminalbench hard27.3%46.2%

Benchmark data from Artificial Analysis.