Introducing Claude Haiku 4.5
Key Takeaways
- Claude Haiku 4.5, a new small model, is now generally available, offering significant cost savings and speed improvements.
- Haiku 4.5 achieves performance comparable to older frontier models (like Claude Sonnet 4) but at one-third the cost and more than double the speed.
- The model is optimized for real-time, low-latency tasks such as chat assistants, customer service, and responsive pair programming.
- It enables new workflow patterns, such as using the frontier model (Sonnet 4.5) to orchestrate subtasks executed by multiple Haiku 4.5 instances.
- Haiku 4.5 shows strong coding performance, achieving 90% of Sonnet 4.5's performance in agentic coding evaluations while being substantially faster and cheaper.
Anthropic announced the immediate availability of Claude Haiku 4.5, a new small model that significantly improves upon the cost-performance ratio in AI. Haiku 4.5 delivers coding performance similar to Claude Sonnet 4 (a model from five months ago) but at one-third the cost and more than twice the speed, even surpassing it in certain tasks like computer usage. This speed and efficiency make it ideal for low-latency applications such as customer service agents and pair programming, and it enhances tools like Claude for Chrome. The model integrates well with the frontier model, Claude Sonnet 4.5, allowing Sonnet to orchestrate complex tasks broken down and executed by multiple Haiku 4.5 instances. Developers can access it via the Claude API with new pricing, and early adopters praise its ability to blend near-frontier quality with blazing speed and cost efficiency, signaling a major shift in the capabilities of smaller models. Safety evaluations indicate that Haiku 4.5 is also substantially more aligned than its predecessor, Claude Haiku 3.5.
