- Opsway
- Posts
- Anthropic’s Anticipated Move Has Arrived
Anthropic’s Anticipated Move Has Arrived
Anthropic’s Claude 3.7 Sonnet is here with hybrid reasoning, excelling in coding (70.3% SWE-bench) and more. Plus, Claude Code for devs! Launched Feb 24, 2025.
Following its outstanding performance in coding, Anthropic, the top choice for coding enthusiasts, has made an eagerly awaited and significant move. This step once again underscores Anthropic’s lasting dominance in the field.
On February 24, 2025, Anthropic introduced its most advanced AI model, Claude 3.7 Sonnet. This model stands out with an industry-first “hybrid reasoning” feature, capable of delivering both fast, instant responses and step-by-step, in-depth analysis for complex problems within a single model. Users can choose between “normal” and “extended reasoning” modes, offering flexibility in speed, cost, and response quality. It excels particularly in areas like coding, mathematics, physics, and instruction-following.

Figure 1: Claude 3.7's Benchmarking Results
Additionally, Anthropic launched “Claude Code,” a tool for developers, in a limited research preview. This tool allows users to delegate coding tasks directly to Claude via the terminal, supporting functions like code reading, editing, test running, and even committing to GitHub. Claude 3.7 Sonnet is available across all Claude plans (including free tiers), though the extended reasoning mode is exclusive to paid plans.
The model’s benchmark results are impressive, achieving a 70.3% success rate on SWE-bench Verified, securing its leadership in coding. It also brings improvements in safety and transparency, with the reasoning process made visible to users. Anthropic aims to enhance human capabilities more effectively with this model, having tested it by playing Pokémon Red and earning three badges! If you want to hear the details of Claude 3.7 from the developers themselves, I recommend watching the video below.
Reply