The new Claude 4 models from Anthropic promise the largest AI -brain ever
- Advertisement -
- Anthropic has two new Claude AI models called Claude Opus 4 and Claude Sonnet 4 debuted
- Claude Opus 4 claims to be the best coding AI in the world
- Claude Sonnet 4 is a smaller, streamlined model with important upgrades from Sonnet 3.7 version.
Anthropic has unveiled Claude 4, the newest generation of his AI models. The company claims that the new Claude Opus 4 and Claude Sonnet 4 models are at the top of the game for AI assistants with unparalleled coding skills and the possibility of functioning independently for a long time.
Claude Sonnet 4 is the smaller model, but it is still an important upgrade of the power of the earlier sonnet 3.7. Anthropic Claims Sonnet 4 is much better in following instructions and coding. It has even been taken over by Github to provide a new Coding agent with Copilot with power. It is probably used much wider, simply because it is the standard model on the free layer for the claude chatbot.
Claude Opus 4 is the flagship model for anthropic and supposedly the best coding AI there is. It can also deal with persistent tasks of several hours by breaking them in thousands of steps to fulfill. Opus 4 also includes the “Extended Thinking” function Anthropic tested on earlier models. Thanks to long -term thinking, the model can pause in the middle of responding to a prompt and use search engines and other tools until it has more data and can resume where it had gone.
That means much more than just longer answers. Developers can train Opus 4 to use all kinds of third -party tools. Opus 4 can even play video games pretty well, showing anthropic how the AI performs during a game by Pokémon Red when access to file and permission is given to build your own navigation guide.
Claude 4 Power
Both Claude 4 models have improved functions centered around tool use and memory. Opus 4 and Sonnet 4 can use and switch tools in parallel between reasoning and search. And their memorial system can store and extract important facts over time when they have access to external files. You do not have to explain what you want with every third prompt.
To ensure that the AI does what you want, but does not overwhelm you with every detail, the models of Claude 4 also offer what the ‘thinking overview’ calls. Instead of a wall of text that describes each of the potential thousands of steps that have been taken to complete a prompt, Claude uses a smaller, secondary AI model to condense the line of thought into something digestible.
A side benefit of the way the new models work is that they are less likely to cheat to save time and processing power. Anthropic said they have reduced quick coupling behavior in tasks that tempt AIS to make their way to a solution (or just come up with something).
The bigger picture? Anthropic clearly fails for the head in AI tools, in particular in coding and agent, independent tasks. Chatgpt and Google Gemini have a larger user base, but anthropic has the means to lure at least a few AI Chatbot users to Claude. With Sonnet 4 available for free users and Opus 4 bundled in Claude Pro, Max, Team and Enterprise plans, Anthropic tries to address both the budget-friendly and the premium AI fans.
Maybe you like it too
- Advertisement -
