Anthropic has released Claude Sonnet 4.6, a model that amounts to a seismic repricing event for the AI industry. It delivers near-flagship intelligence at mid-tier cost, landing squarely in the middle of an unprecedented corporate rush to deploy AI agents and automated coding tools. # The model is a full upgrade across coding, computer use, long-context reasoning, agent planning, knowledge work, and design. It features a 1M token context window in beta, and is now the default model in [Claude.aihttps://claude.ai){rel=“nofollow”} and Claude Cowork. Here’s the kicker: pricing holds steady at $3/$15 per million tokens — the same as its predecessor, Sonnet 4.5. ## The $5x Cost Revolution That pricing detail is the headline that matters most. Anthropic’s flagship Opus models cost $15/$75 per million tokens — five times the Sonnet price. Yet performance that would have previously required reaching for an Opus-class model is now available with Sonnet 4.6. For the thousands of enterprises deploying AI agents that make millions of API calls per day, that math changes everything. ### Benchmark Breakdown The benchmark table Anthropic released paints a striking picture: | Benchmark | Sonnet 4.6 | Opus 4.6 | |———–|————-|———-| | SWE-bench Verified (coding) | 79.6% | 80.8% | | OSWorld-Verified (computer use) | 72.5% | 72.7% | | GDPval-AA (office tasks) | 1633 | 1606 | | Agentic Financial Analysis | 63.3% | 60.1% | On office tasks, Sonnet 4.6 actually surpassed Opus 4.6. On agentic financial analysis, it hit 63.3% beating every model in the comparison. ## Computer Use: From Experimental to Near-Human in 16 Months One of the most dramatic storylines is Anthropic’s computer use journey. Sonnet 4.6 scored 72.5% on OSWorld-Verified, up from just 14.9% when the capability first launched in October 2024. That’s a nearly 5x improvement in 16 months. This positions Claude as arguably the most capable AI assistant for computer-based tasks — a critical capability as enterprises build agents that autonomously navigate browsers, execute code, and interact with enterprise software. ## Claude Code Preference In [Claude Codehttps://claude.com/product/claude-code){rel=“nofollow”}, early testing found that users preferred Sonnet 4.6 over Sonnet 4.5 roughly 70% of the time. Even more striking: users preferred Sonnet 4.6 to Opus 4.5 59% of the time. Users reported: - Significantly less prone to over-engineering and “laziness” - Meaningfully better at instruction following - Fewer false claims of success and hallucinations - More consistent follow-through on multi-step tasks ## Why It Matters For enterprises, the calculus has fundamentally shifted: 1. Scale Economics: Run AI agents at 1/5th the cost without sacrificing quality 2. Agentic Workloads: Match Opus performance on computer use and coding at Sonnet prices 3. Default Choice: Sonnet 4.6 is now the default — most users won’t need Opus 4. Stable Pricing: Despite massive improvements, Anthropic held the line on pricing The release is available now via the [Anthropic APIhttps://www.anthropic.com/api){rel=“nofollow”} and [Claude.aihttps://claude.ai){rel=“nofollow”}.