AI startup Anthropic has officially entered the top-tier AI race with the launch of its new model family, Claude 3. The announcement on Monday positions the company as a direct and formidable competitor to industry leaders like OpenAI and Google, with Anthropic claiming its most powerful model surpasses the performance of GPT-4.
The Claude 3 family consists of three distinct models, designed to offer users a balance of intelligence, speed, and cost. The models are Haiku, Sonnet, and the flagship model, Opus. According to benchmark data released by Anthropic, Opus has outperformed OpenAI’s GPT-4 on key industry evaluations, including undergraduate-level knowledge (MMLU) and graduate-level reasoning (GPQA). This marks a significant milestone, as GPT-4 has largely been the industry’s performance standard since its release.
A major upgrade for the Claude series is the introduction of multimodal capabilities. All Claude 3 models can now analyze and process both text and visual inputs, such as photos, charts, and diagrams. This brings Anthropic’s offerings in line with competitors like Google’s Gemini and GPT-4 with Vision, allowing for more complex and context-aware applications. The company highlighted that this feature was a primary request from its enterprise user base.
Anthropic stated that the mid-tier Sonnet model is now powering the free experience on its chat interface, claude.ai, with the superior Opus model available to Pro subscribers. Both Sonnet and Opus are also accessible via API in major regions. The fastest and most compact model, Haiku, is slated for release soon.
This launch intensifies the competition among leading AI labs, signaling that the rapid pace of innovation is not slowing. For businesses and developers, the arrival of a competitive alternative to GPT-4 provides more options and could drive down costs while pushing the boundaries of what AI can achieve.
English 
Deutsch