Claude Opus 4.8 Review: Better At What’s It Good At, Worse At What It’s Not
Claude Opus 4.8 is a mixed update: it improves on math, coding, and some reasoning, while creative writing stays roughly flat or slightly worse than 4.7. In tests, it produced a vivid but only competent time-travel story, and its prose lacked the momentum of competing models. Its coding performance was the strongest result: it built a solid typing-zombie game, caught and fixed its own bugs, and handled multi-shot improvements well. It also solved a difficult FrontierMath problem correctly, a clear gain over 4.7. But it failed a whodunit by constructing a persuasive yet wrong answer, and it collapsed on a 300K-token haystack while the smaller one passed but then refused to answer after finding the planted needles. The biggest drawback is cost: token usage is so high that practical use quickly exhausts quotas, making it more attractive for high-paying coders than for general users.
