Anthropic’s New AI Model: Claude 3.5 Haiku
Anthropic’s newest AI model, Claude 3.5 Haiku, has been introduced. However, it comes at a higher price compared to the previous generation and does not yet have the capability to analyze images.
Claude 3.5 Haiku, unveiled by Anthropic last month, demonstrates performance that matches or surpasses their flagship model, Claude 3 Opus, on specific benchmarks. Accessible through Anthropic’s API and various third-party platforms like AWS Bedrock, Claude 3.5 Haiku is designed for tasks such as coding suggestions, data extraction, labeling, and content moderation according to Anthropic.
Previously, Anthropic indicated that Claude 3.5 Haiku would be priced similarly to its predecessor, Claude 3 Haiku. However, this has changed.
“During final testing, Haiku outperformed Claude 3 Opus, our previous flagship model, on multiple benchmarks — at a lower cost. Consequently, we have adjusted the pricing for Claude 3.5 Haiku to reflect its enhanced intelligence,” Anthropic stated in a post on X.
The pricing for Claude 3.5 Haiku now starts at $1 per million input tokens (~750,000 words) and $5 per million output tokens, compared to 25 cents per million input tokens and $1.25 per million output tokens for Claude 3 Haiku — a 4x increase.
Although Claude 3.5 Haiku lacks image analysis capabilities at launch, it can generate longer pieces of text than Claude 3 Haiku and has a more recent knowledge cutoff, allowing it to reference more recent events. Nevertheless, the decision to raise the cost of a model within a series is uncommon in the AI industry and raises concerns about Anthropic’s pricing strategy going forward.
