Claude Haiku 4.5 is Anthropic's newest miniature artificial intelligence system that matches the coding efficiency of Sonnet 4 at one-third the expense and double the pace, making it perfect for real-time, budget-conscious tasks.
Efficiency with High PerformanceMatches Claude Sonnet 4's performance without exceeding its costs, which are significantly lower: priced at just $0.003 per million input tokens and $0.0167 per million output tokens.
Enhanced Context AnalysisSupports a 200,000-token context window with up to 64,000 output tokens, capable of handling both text and images.
Improved Security MeasuresUnder AI Safety Level 2 (ASL-2), our model has lower misalignment rates compared to Sonnet 4.5 and Opus 4.1, making it the most safe model we have developed so far.
Capability of Multiple AgentsCan function as subordinate agents under Sonnet 4.5 for concurrent task execution and intricate problem-solving
