Скачать или смотреть Introducing Claude Sonnet 4.5: The Best Coding Model and Leap in Frontier Intelligence

This video introduces Claude Sonnet 4.5, which is described as the best coding model in the world. It is also the strongest model for building complex agents and the best model at using computers. Claude Sonnet 4.5 shows substantial gains in reasoning and math and represents Anthropic's most powerful model to date.

Key updates and capabilities highlighted include:
• State-of-the-Art Performance: Sonnet 4.5 is state-of-the-art on the SWE-bench Verified evaluation, which measures real-world software coding abilities. It has demonstrated the ability to maintain focus for more than 30 hours on complex, multi-step tasks.
• Computer Use: The model shows significant advancements in computer use capabilities, now leading the OSWorld benchmark at 61.4%. A demo showcases the Claude for Chrome extension, where Claude works directly in a browser, navigating sites, filling spreadsheets, and completing tasks.
• Product Upgrades: The release includes major upgrades to Anthropic's products, such as the introduction of checkpoints in Claude Code, a refreshed terminal interface, and a native VS Code extension.
• Claude Agent SDK: Developers can now access the infrastructure that powers Anthropic’s frontier products through the new Claude Agent SDK. This SDK shows impressive benefits for a wide variety of tasks, not just coding.
• Alignment: Claude Sonnet 4.5 is the most aligned frontier model released yet, showing large improvements across several areas of alignment, including reducing concerning behaviors like sycophancy and deception.
• Research Preview: A temporary research preview, "Imagine with Claude," is also featured, which demonstrates the model generating software on the fly without prewritten code.


#ClaudeSonnet45 #Anthropic #AImodel #CodingModel #FrontierIntelligence #Agents #AgenticCoding #SoftwareDevelopment #ComputerUse #Reasoning #Math #SWEbenchVerified #OSWorld #ClaudeCode #ClaudeAgentSDK #ImagineWithClaude #VSCodeExtension #Alignment #Safety #SecurityAgents #Litigation #Finance