Hasty Briefsbeta

Bilingual

Claude Sonnet 4.6

5 hours ago
  • #AI
  • #Claude Sonnet
  • #Machine Learning
  • Claude Sonnet 4.6 is the most capable Sonnet model yet, with upgrades in coding, computer use, long-context reasoning, and more.
  • Features a 1M token context window in beta and is now the default model for Free and Pro plans on claude.ai and Claude Cowork.
  • Pricing remains the same as Sonnet 4.5, starting at $3/$15 per million tokens.
  • Improved coding skills, consistency, and instruction following make Sonnet 4.6 preferred by developers over its predecessor and even Claude Opus 4.5 in some cases.
  • Major improvements in computer use skills, with human-level capability in tasks like navigating complex spreadsheets or filling out multi-step web forms.
  • Sonnet 4.6 shows significant resistance to prompt injection attacks, performing similarly to Opus 4.6 in safety evaluations.
  • Performance improvements across benchmarks, approaching Opus-level intelligence at a more practical price point.
  • Early testing shows users prefer Sonnet 4.6 over Sonnet 4.5 70% of the time in Claude Code, with better context reading and logic consolidation.
  • Sonnet 4.6's 1M token context window allows for effective reasoning across entire codebases, lengthy contracts, or dozens of research papers.
  • Notable improvements in frontend code, financial analysis, visual outputs, and design sensibility, requiring fewer iterations for production-quality results.
  • Matches Opus 4.6 performance on OfficeQA, a significant upgrade for document comprehension workloads.
  • Excels at complex code fixes, bug detection, and agentic coding at scale, with strong resolution rates and consistency.
  • First Sonnet model to offer frontier-level reasoning in a smaller, more cost-effective form factor.
  • Significant improvements in answer retrieval, with better recall on specific workflows in the Financial Services Benchmark.
  • Outperforms Sonnet 4.5 in heavy reasoning Q&A by 15 percentage points in evaluations by Box.
  • Achieves 94% on the insurance benchmark, the highest-performing model for computer use in mission-critical workflows.
  • Delivers frontier-level results on complex app builds and bug-fixing, becoming the go-to for deep codebase work.
  • Produces the best iOS code tested for Rakuten AI, with better spec compliance, architecture, and modern tooling.
  • Strong performance on branched and multi-step tasks like contract routing, conditional template selection, and CRM coordination.
  • Perfect design taste in building frontend pages and data reports, requiring less hand-holding.
  • Exceptionally responsive to direction, delivering precise figures and structured comparisons while generating useful ideas.
  • Supports adaptive thinking, extended thinking, and context compaction in beta on the Claude Developer Platform.
  • API updates include automatic code execution for filtering search results, improving response quality and token efficiency.
  • Opus 4.6 remains the strongest option for tasks requiring the deepest reasoning, such as codebase refactoring and coordinating multiple agents.
  • Claude in Excel now supports MCP connectors, allowing integration with tools like S&P Global, LSEG, and FactSet.
  • Available on all Claude plans, Claude Cowork, Claude Code, API, and major cloud platforms, with the free tier upgraded to Sonnet 4.6 by default.