AB
AB
AiBoss

All Categories

project

Claude 4 - Anthropic's latest AI programming series model

Claude 4 is a next-generation AI model from Anthropic, comprising Claude Opus 4 and Claude Sonnet 4. Opus 4 is currently the world's most powerful programming model, excelling at complex tasks and long-running operations...

What is Claude 4?

Claude 4 is a next-generation AI model from Anthropic, comprising Claude Opus 4 and Claude Sonnet 4. Opus 4 is currently the world's most powerful programming model, excelling at complex tasks and long-running workflows such as code generation, optimization, and debugging. Claude Sonnet 4 offers significant improvements in programming and reasoning capabilities, with more accurate responses, making it suitable for everyday use. Both support instant response and deep thinking modes, allowing for parallel tool usage and significantly enhanced memory capabilities. Claude 4 introduces tool-assisted extended thinking and memory file management features, further improving the practicality and efficiency of the AI Agent.

Claude 4's main functions

  • Code generation and optimizationClaude Opus 4 is a top-tier programming model, leading in scores on SWE-bench and Terminal-bench, and capable of generating high-quality code.
  • Long task processingClaude Opus 4 can continuously process complex and long tasks, working for several hours, which is significantly better than other models.
  • Code editing and debuggingClaude Sonnet 4 excels in code editing and debugging, enabling precise modification of code across multiple files.
  • Advanced reasoning abilityClaude Opus 4 can solve complex problems and handle tasks that other models cannot.
  • Multimodal capabilitiesClaude 4 performs well in coding, reasoning, multimodal, and agent tasks.
  • Tool usage and expanding thinkingClaude 4 can use tools (such as web searches) to expand its thinking and improve response quality. The model can use tools in parallel to improve task processing efficiency.
  • Local file access and memory capabilitiesAfter developers grant local file access permissions, the model can extract and save key information, improving task continuity and performance.
  • Reduce shortcut behaviorClaude 4 reduces the use of shortcuts or vulnerabilities by 65% compared to Sonnet 3.7 when performing tasks.
  • Improved memoryClaude Opus 4 can create and maintain "memory files" that store critical information, improving awareness and consistency in long-term tasks. For example, it can create a navigation guide when playing a Pokémon game.
  • Reflection and SummaryClaude 4 introduces a summary function to streamline lengthy thinking processes, requiring it to be used only in about 5% of cases.

Claude 4's test performance

  • Claude Opus 4:
    • SWE-benchClaude Opus 4 scored 72.5% in the SWE-bench test, significantly outperforming other models.
    • Terminal-benchThe Claude Opus 4 scored 43.2% in the Terminal-bench test, demonstrating excellent performance.
  • Claude Sonnet 4 :
    • SWE-bench Claude Sonnet 4 achieves an excellent coding efficiency of 72.7% on SWE-bench.

Claude 4 product pricing

  • Claude Opus 4$15 per million tokens input, $75 per million tokens output.
  • Claude Sonnet 4$3 per million tokens input, $15 per million tokens output.
  • Subscription PlanUsers who subscribe to the Pro, Max, Team, and Enterprise plans will have access to and expanded thinking for Claude Opus 4 and Claude Sonnet 4, with Sonnet 4 available to free users.

Claude 4's project address

Application scenarios of Claude 4

  • Programming aidsQuickly generate and optimize code to improve development efficiency.
  • AI Agent: To perform complex tasks, call external tools, and maintain contextual consistency.
  • Software developmentProvide code suggestions within the IDE to simplify the review process.
  • Data Analysis and ProcessingGenerate data visualization code to process and analyze data.
  • Natural Language ProcessingGenerates high-quality text and supports multilingual translation.
Back