Skip to main content

Claude Haiku 4.5

Large Language Model
Visit Website

Anthropic's fastest and most compact model.

Developer

Anthropic

Release Date

October 01, 2025

Pricing

Paid

Key Features

Fast Response
Text Generation
Code Writing
Efficient Processing

Use Cases

Real-time Applications

Perfect for real-time applications applications

Quick Responses

Perfect for quick responses applications

Cost-effective Tasks

Perfect for cost-effective tasks applications

High-volume Processing

Perfect for high-volume processing applications

What is Claude Haiku 4.5?

Claude Haiku 4.5 is Anthropic's fastest and most affordable model, released on October 15, 2025. It delivers performance comparable to Claude Sonnet 4 — which was Anthropic's top model just two months before — at a third of the cost. At $1 per million input tokens and $5 per million output tokens, it's the go-to choice for high-volume applications, latency-sensitive workloads, and cost-constrained deployments.

How Good is Claude Haiku 4.5?

Haiku 4.5 scores 73.3% on SWE-bench Verified — only 3.9 points below Sonnet 4.5 (77.2%) while costing three times less. For autonomous coding tasks, it delivers approximately 90% of Sonnet 4.5's performance. This makes the decision straightforward: if your task doesn't require maximum accuracy, Haiku 4.5 saves significant money at scale.

It generates around 97 tokens per second — 83% faster than Sonnet 4.6 and 116% faster than Opus 4.6. For customer-facing applications where response latency matters, Haiku 4.5 is the clear choice. It supports up to 64,000 output tokens per response, a major improvement over its predecessor's 8,192 token limit.

What's New in Haiku 4.5?

This is the first Haiku model to include extended thinking, computer use, and context awareness — capabilities previously exclusive to Sonnet and Opus models. Extended thinking allows the model to reason through complex problems before responding. Computer use enables autonomous interaction with software interfaces. These additions dramatically expand what's possible at Haiku pricing.

Claude Haiku 4.5 Pricing

Input: $1 per million tokens. Output: $5 per million tokens. Batch processing offers a 50% output discount ($2.50/M) for asynchronous workloads. Prompt caching writes: $1.25/M, reads: $0.10/M. For a customer support bot handling 100,000 conversations monthly with 1,000 input and 500 output tokens each, total cost is $125/month — extremely cost-effective.

Claude Haiku 4.5 vs Competitors

Against GPT-5 mini ($0.30/$1.20 per million tokens), Haiku 4.5 costs more but offers stronger reasoning and Anthropic's reliability. Against DeepSeek V3.2 ($0.25/$0.38), Haiku 4.5 is pricier but has better instruction following and Western data center hosting. For teams already in Anthropic's ecosystem, Haiku 4.5 is the efficient default for simple tasks.

Frequently Asked Questions

Is Claude Haiku 4.5 good for coding?

Yes, especially for simple to medium complexity tasks. It scores 73.3% on SWE-bench Verified. For complex architecture decisions or large codebase analysis, upgrade to Sonnet 4.6 or Opus 4.6.

What is the context window for Haiku 4.5?

200,000 tokens — the same as Sonnet models. This is unusual: Haiku 4.5 matches flagship models on context capacity but costs 5x less than Opus.

API Available

Integrate Claude Haiku 4.5 into your applications

Similar AI Models

GPT-4.1

OpenAI's smartest non-reasoning model with enhanced capabilities...

Large Language Model Learn More →

Claude Sonnet 4.5

Anthropic's smartest and most efficient model for everyday use...

Large Language Model Learn More →

GPT-5

OpenAI's latest flagship model series with advanced reasoning....

Large Language Model Learn More →

Perplexity Ai

Perplexity AI is an AI-powered answer engine that combines generative AI with real-time web search t...

Large Language Model Learn More →

Kimi 2

Your all-in-one AI assistant - now with K2 Thinking, the best open-source reasoning model. Solves ma...

Large Language Model Learn More →

Claude Opus 4.1

Anthropic's most powerful model for complex tasks....

Large Language Model Learn More →