LaunchAI Models
19 days ago
Claude Opus 4.8 launched with minor improvements, stress on honesty
Opus 4.8 costs the same as its predecessor ($5M input, $25M output). A key improvement is in honesty: it is four times less likely to leave flaws in code unnoticed, and achieves the lowest incorrect rate on benchmarks by abstaining when uncertain.
·
19 days ago