Anthropic says its new AI model “maintained focus” for 30 hours on multistep tasks

Anthropic has released Claude Sonnet 4.5, an AI language model positioned as its most capable to date, boasting significant advancements in coding, computer use, and sustained multi-step task execution for over 30 hours. This launch, accompanied by Claude Code 2.0 and the Claude Agent SDK, addresses critical limitations in long-duration AI agent coherence and strengthens Anthropic's competitive stance, particularly in developer-focused AI applications, signaling potential for broader enterprise adoption and market share gains in the evolving AI sector.

Analysis

Anthropic has launched Claude Sonnet 4.5, positioning it as its most capable model, with specific claims of market-leading coding capabilities and enhanced reasoning. A key technical milestone highlighted is the model's ability to operate on complex, multi-step tasks for over 30 hours, directly addressing the common issue of coherence degradation in long-duration AI agents. This release is strategically significant as Sonnet occupies the mid-range of Anthropic's offerings, aiming for an optimal balance of performance and cost, which is a critical factor for enterprise-scale adoption. The simultaneous launch of Claude Code 2.0 and a developer-focused Agent SDK underscores a targeted strategy to deepen its footprint within the software development community. While Anthropic is a private entity, these advancements set a new competitive benchmark for agentic performance and developer tools, challenging publicly traded rivals in the AI sector to demonstrate comparable capabilities.

Key Decisions for Investors

• Given Anthropic is private, investors with exposure to publicly traded AI leaders should closely monitor their competitive responses to the new coding and agentic benchmarks set by Claude Sonnet 4.5.
• The emphasis on cost-performance and extended task duration signals a push towards enterprise-grade applications; therefore, watch for any announcements of enterprise adoption or partnerships as a key indicator of the commercial viability of advanced AI agents.
• The launch of the Agent SDK is a strategic move to build a developer ecosystem, so investors should assess the strength and adoption rates of developer tools offered by competing public AI and cloud platforms, as this is a crucial factor for long-term market share.