Anthropic releases Claude 4 models with enhanced coding and reasoning capabilities

OpenAI rival Anthropic has officially launched Claude 4, marking a significant leap forward in artificial intelligence (AI) capabilities with two powerful new models: Claude Opus 4 and Claude Sonnet 4. This latest generation sets new benchmarks for coding excellence, advanced reasoning, and AI agent performance, promising to transform how developers and businesses interact with AI. 

[Photo/Anthropic]

According to Anthropic, Claude Opus 4 represents what the company calls the world's most advanced coding model, demonstrating exceptional performance on complex, long-running tasks that require sustained focus over several hours. Anthropic reports that Opus 4 can maintain consistent performance across thousands of steps, potentially enabling it to tackle sophisticated coding challenges that previously required human expertise.

Claude Sonnet 4 represents a substantial upgrade from its predecessor, Sonnet 3.7, while maintaining the efficiency that made it popular among developers. With an impressive 72.7% score on SWE-bench, Sonnet 4 delivers enhanced coding capabilities and improved instruction-following, making it ideal for everyday development tasks. 

[Photo/Anthropic]

Anthropic claims that both models introduce hybrid reasoning capabilities, offering users a choice between near-instant responses for quick queries and extended thinking modes for complex problem-solving. According to the company, this flexibility allows the AI to alternate between rapid tool use and deep reasoning, potentially improving response quality for challenging tasks.

Both models are now available across multiple subscription plans. Anthropic offers its Pro plan at $20 per month, Max plan at $60 per month, and Teams plan at $25 per month per user, while continuing to provide Claude Sonnet 4 access through a free version.