推文

@zouyanjian · 2026-10-12 21:11

At $0.10 per 1M input tokens for prompts up to 100K, Haiku 5.5 makes cheap subagents tempting. Its Terminal-Bench 4.0 score is 39.2% vs Sonnet 5.5’s 70.6%. What would you trust Haiku to handle?

曝光 111 · 评论 1 · 点赞 4 · 书签 0 · 曝光/时 84.7580991276892