Anthropic's Claude has overtaken ChatGPT as the top-ranked paid AI assistant in South Korea, according to mobile app store data and third-party analytics released this week. The reversal marks a notable shift in a market that has historically favored OpenAI's products, driven by South Korean enterprise customers who cite Claude's stronger handling of formal Korean-language documents and its superior performance on multi-step reasoning tasks common in legal and financial services. Claude's ascent in South Korea is not occurring in isolation: the company has seen similar upticks in Japan and Taiwan, where enterprise procurement teams increasingly evaluate AI assistants on regional language quality rather than English-language benchmarks alone.
Anthropic simultaneously acknowledged a set of performance reliability issues that have affected a subset of Claude users over the past three weeks. Engineers traced the root cause to load-balancing irregularities introduced during a capacity expansion intended to support the model's global growth. The company pushed a patch that reduced error rates by approximately 74 percent within 48 hours, though some users reported continued intermittent delays on complex multi-turn conversations. Anthropic's transparency in publishing a public post-mortem was itself notable — a departure from the industry's typical practice of silently resolving infrastructure incidents without explanation.
The juxtaposition of market momentum and reliability turbulence illustrates the operational complexity of scaling a frontier AI system. South Korean enterprise wins are commercially meaningful: South Korean companies tend to commit to multi-year software contracts and integrate tools deeply into internal workflows. Winning procurement decisions now creates compounding switching costs. However, reliability gaps can rapidly erode hard-won trust, particularly in regulated industries where AI downtime carries direct operational consequences. Anthropic's engineering response and post-mortem transparency suggest the company understands the stakes — and that its current challenge is scaling infrastructure at a pace that matches its market momentum.