Anthropic introduces Claude Sonnet 5: AI agents become more affordable

Anthropic, one of the leading laboratories in the field of artificial intelligence, has officially announced its new Claude Sonnet 5 model. Despite being the company's mid-tier development, it is impressing many with its agentic capabilities and level of autonomous task execution. Users can now access features previously exclusive to the most expensive and largest models at significantly lower prices. This is reported by Techcrunch.com reports .

Anthropic representatives state that Claude Sonnet 5 not only works with text but also has the potential to plan, use tools like browsers and terminals, and perform complex tasks without human intervention. Such "agentic" features are becoming a new standard in the AI market today. In this direction, OpenAI is competing with its GPT-5.6 Sol model, and Google with its Gemini 3.5 Flash models.

A balance of high performance and cost-efficiency

The main advantage of the Claude Sonnet 5 model lies in its price-to-performance ratio. According to data provided by Anthropic, the new model shows results close to its most powerful and expensive predecessor, Opus 4.8. For example, in tests conducted on software code writing, Sonnet 5 recorded a 63.2% result, while the top-tier Opus 4.8 indicator is 69.2%.

Notably, in some knowledge-based benchmarks, Sonnet 5 even managed to outperform the Opus 4.8 model. This allows developers and business representatives to significantly reduce the costs of using high-quality AI services. The company offers users various models to maintain a balance of price and quality based on their needs.

Regarding pricing, until August 31, the price for Claude Sonnet 5 is set at $2 per million input tokens and $10 per million output tokens. Starting in September, the price for input tokens will rise to $3. Nevertheless, this figure is cheaper than many professional models from OpenAI and Google.

Ability to make independent decisions

Another unique aspect of the new model is its self-correction feature. According to experts who participated in the tests, Claude Sonnet 5 does not stop when performing complex tasks like previous versions, but independently reviews the result and corrects its errors.

For developers and technology enthusiasts in Uzbekistan, this news may further popularize the use of AI tools. Especially for startups and small business projects, using cheap but powerful AI agents will be an important factor in increasing productivity. Starting today, Claude Sonnet 5 is being introduced as the primary model for all free and Pro users.