Anthropic recently introduced Claude Opus 4.6, which excels in long-context retrieval. Now, it has also updated Claude Sonnet, the company’s faster, more cost-effective model.
Sonnet 4.6 is the most capable Sonnet model yet, with improved capabilities in coding, computer use, long-context reasoning, agent planning, knowledge work, and design. While it doesn’t outperform Opus in all aspects, each model has its own strengths that are consistently growing.
What’s New with Sonnet 4.6?
Claude Sonnet 4.6 improved in consistency and instruction following, specifically for coding skills. It can now better perform real-world office tasks, making it a huge step up from previous Sonnet models and comparable to previous Opus models.
Developers who had early access to the new Sonnet have said that 4.6 is better than the past model by a wide margin. Some even prefer it over Claude Opus 4.5. Safety evaluations also show that Sonnet 4.6 is improving when it comes to resisting malicious actors.
Over the past 16 months, Sonnet has made huge improvements in computer use. The OSWorld scores (standard benchmarks for AI computer use) increased gradually over time. Back in October 2024, Sonnet 3.5 only scored 14.9. In October 2025, Sonnet 4.5 jumped up to 61.4. Now, in February 2026, Sonnet 4.6 has achieved 72.5.
How Does it Compare to Other AI Models?
Sonnet 4.6 has improved on every benchmark since 4.5. While Opus 4.6 excels in agentic terminal coding, agentic computer use, agentic tool use, and agentic search, Sonnet proves to be the best in other categories.
The new Sonnet model scored the highest in agentic financial analysis and office tasks, outranking Opus 4.6, Gemini 3 Pro, and ChatGPT-5.2. Overall, it’s getting close to Opus-level intelligence while remaining at a price that’s more practical for the average user.
Users have particularly noted lots of broad improvements for Sonnet 4.6. Frontend code, financial analysis, and polished visual outputs are some of the most commonly mentioned improvements.
Using Sonnet 4.6
Sonnet 4.6 is now available in all Claude subscription plans, and it’s used by default in the free tier. It can use both adaptive thinking and extensive thinking, depending on which you prefer. Adaptive thinking decides how deeply to think based on the context while extended thinking focuses on deep, multi-step reasoning.
However, Opus 4.6 is still the recommended option for deep reasoning tasks, such as codebase refactoring or coordinating multiple agents in a workflow.
One aspect that users can enjoy on all Claude models is an ad-free experience. As part of Anthropic’s Super Bowl commercials, the company announced that they won’t be incorporating ads into their AI chats because they want to keep responses genuine.
Get Your Brand Noticed in AI Searches
AI keeps evolving, offering models for all types of tasks. If you want your brand to keep showing up everywhere your audience is, don’t forget about AI optimization. Appearing in AI searches is the best way to help more customers discover your brand.
Contact Avenue Z today to experience how we drive revenue and reputation for brands.
We are the Agency for Influence
Discover new ways to drive revenue and build reputation for your brand.
