Claude 3.5 Haiku
Provider: Anthropic US
Overview
Anthropic's Claude 3.5 Haiku is a cost-optimized model designed for high-volume, lightweight tasks. It offers a balance of price and performance within the Haiku family, targeting applications requiring a broad context window and multimodal capabilities.
Strengths
- Offers an extremely large 200,000 token context window for extensive document processing.
- Includes vision capabilities, enabling multimodal input understanding and analysis.
- Supports function-calling, allowing integration with external tools and APIs for complex workflows.
- Provides a very competitive cache read price of $0.08/1M tokens for repeat requests.
Best use cases
- Analyzing and summarizing extensive legal, technical, or research documents efficiently.
- Generating descriptions or insights from images, charts, and diagrams.
- Building intelligent agents that interact with external services via function-calling.
- Powering customer support chatbots requiring a broad conversational history and context.
- High-volume data extraction or classification tasks where cost efficiency is key.
Pricing position
Claude 3.5 Haiku is positioned as a mid-range cost-optimized model within Anthropic's lineup, priced at approximately $0.80/M tokens for input and $4/M tokens for output. It is notably more expensive than its sibling Claude 3 Haiku ($0.25/M in, $1.25/M out) but less costly than Claude Haiku 4.5 ($1/M in, $5/M out) and the Sonnet series. Compared to similarly-priced peers, its pricing is competitive with Alibaba's Qwen3 Max models and OpenAI's GPT-5.4 Mini, while models like Kimi K2.6 ($0.73/M in, $3.49/M out) and Deepseek's R1 ($0.7/M in, $2.5/M out) offer lower per-token costs.
Caveats
This model does not offer a free tier, meaning developers will incur costs from the first token of use. While its 200,000 token context window is substantial, it does not match the 1,000,000 token context available with Anthropic's more expensive Sonnet models. For use cases strictly prioritizing the lowest possible cost, models like Claude 3 Haiku or Deepseek's R1 may offer a more budget-friendly alternative.
Notes drafted by an LLM from the figures on this page (2026-05-21) and rewritten when the price changes.
Key facts
- Max output
- 8K tokens
- Tracked since
- 2026-05-08
- Last updated
- 2026-06-22
Features
Price history
- 2026-06-22 No longer listed
Monthly cost estimate
| Scale | Input (M) | Output (M) | Monthly |
|---|---|---|---|
| 0.1M (hobby) | 0.1 | 0.05 | $0.28 |
| 1M (small app) | 1 | 0.5 | $2.80 |
| 10M (mid scale) | 10 | 5 | $28.00 |
| 100M (large) | 100 | 50 | $280.00 |
※ Cache and batch discounts not factored in. Detailed calculator
Official resources
Other models from Anthropic
Similar-priced models
Source: OpenRouter API