Jiakai's Blog — Page 2

Reviews Sonnet 4.5 for vibe coding, writing, Imagine with Claude, and Pro quotas, while weighing capability gains against Anthropic’s tighter limits.

Tests a Claude Max subscription and Claude for Chrome on live browsing and research tasks, including its automation strengths, limits, and account risk.

Explains how Opus 4.1 and Claude Code restored confidence in Claude for writing and coding after disappointing Opus 4 outputs.

Tests free and Basic Manus agents: paid mode improves effort and PPTs, but weak source judgment, opaque models, and bot checks undermine reliability.

Tests Grok 4’s X-native search, agent workflow, low censorship, image recognition, Deep Research, and weaker Chinese-language performance.