Claude 3: Definition and Examples

Claude 3 refers to the third major generation of language models created by Anthropic, a company specializing in AI safety. Launched in March 2024, this family comes in three models — Haiku, Sonnet, and Opus — which differ in reasoning power, execution speed, and pricing. This multi-tier architecture allows users to choose the model best suited to their use case.

Claude 3 marked a turning point for Anthropic by introducing multimodal capabilities: all three models can analyze images in addition to text, significantly expanding the range of possible applications. The Claude 3 family also pushed boundaries in terms of context window, with a standard capacity of 200,000 tokens, allowing very long documents to be processed in a single request.

In terms of performance, Claude 3 Opus positioned itself as a direct competitor to GPT-4 on benchmark tests, while Sonnet offered excellent value for money for common professional uses, and Haiku stood out for its speed and low cost for simple tasks. This range allowed Anthropic to establish itself as a major player in the LLM market.

Claude 3 has since been succeeded by the Claude 3.5, Claude 4, and Claude 4.5/4.6 families, but it remains an important reference in the history of generative AI as the moment when Anthropic achieved parity with the best models on the market.