Anthropic's Claude 3 Challenges ChatGPT with Near-Human Performance
• Anthropic claims its new Claude 3 chatbot outperforms ChatGPT and Google's Gemini on key AI benchmarks, even reaching "near-human" performance on some tasks • There are 3 versions - Haiku, Sonnet, and Opus - with Opus being the largest and most powerful, capable of both text and image inputs • Claude 3 models promise near real-time response times and increased accuracy on longer, multi-step instructions • Anthropic Opus scored 14.7% higher than GPT-4 on graduate-level reasoning tests and beat it on math, coding, reasoning, and knowledge tasks • Claude 3 Sonnet and Opus are available now via API, with Haiku coming soon; Sonnet is also accessible through Amazon Bedrock and Google Cloud