Anthropic He has just presented two new versions of Claude, his famous conversational. The most outstanding is Claude Opus 4its most advanced coding model to date and expert in solving complex problems. On the other hand we find Claude Sonnet 4the evolution of the sonnet 3.7 model, which presents great coding and reasoning capabilities, aimed at everyday use.
Both are hybrid modelsthat is, they offer the option to activate or not their advanced reasoning capabilities, depending on the complexity of the task to be performed. It should be noted that Claude Sonnet 3.7, presented in February of this year, was the first hybrid model created by Anthropic.
Claude Opus 4 and Claude Sonnet 4 are already found Available in Pro, Max, Team and Enterprise plans Claude, while the Free plan only includes Sonnet 4. Likewise, both are included in the Anthropic API, Amazon Bedrock and Vertex ai of Google Cloud.
So are the new Claude 4 models
Claude 4 models are The most powerful of the Claude family to dateestablishing new standards in coding, deep and agricultural reasoning. Both IAS have also been optimized to avoid using shortcuts or gaps to complete taskshaving reduced the chances of presenting this behavior by 65%.
It should be noted that these models They can use tools, such as the web search, while applying their extended reasoning. This allows them to alternate between reasoning and use of tools to offer better results and responses. Although, for now, this capacity is still in Beta phase.
As explained by Anthropic, both Claude Opus 4 and Claude Sonnet 4, «they can Use tools in parallel, follow more precision instructions And, when developers give access to local files, they demonstrate significantly improved memory capabilitiesextracting and saving key data to maintain continuity and generate tacit knowledge over time ».
Claude Opus 4
Claude Opus 4 is designed to offer a sustained yield in complex and long -lasting taskswhich demand a concentrated effort and execution of thousands of steps. This AI has the ability to work continuously for hoursmarking a before and after in the family of Claude models.
In addition, it has a remarkably superior memory to that of previous models. Which, added to the aforementioned skills, have helped him improve his game experience when playing Pokemon. An experiment that was already tested with Claude 3.7 Sonnet.
Another prominent aspect is your potential to boost workflows of ia agentsmanaging to significantly expand the ability of these agents. All these characteristics make it a tool aimed at addressing high complexity tasks In various fields, such as programming, research, science or writing, among others.
Anthropic defines Claude Opus 4 as “The best coding model in the world”and to justify his statements he has shared some of the results obtained by the model in different evidence. For example, Claude Opus 4 achieved a 72.5% score in Swe-Bench and 43.2% in Terminal-Bench. 
In addition to this, the developer has also shared the experience of companies like Cognition, cursor, replit, block or rakute, using Claude Opus 4 For various tasks.
Claude Sonnet 4
For its part, Claude Sonnet 4 is less advanced than Claude Opus 4, but it is The most pointer model of the Sonnet range. His abilities of deep reasoning and of coding They are highly precise and presents great improvements in their ability to respond to the instructions that are given.
According to the data presented by Anthropic, Claude Sonnet 4 obtained a result of the 72.7% in Swe-Bench coding tests. «The model balances performance and efficiency for internal and external cases, with greater manageability for greater control over implementations. While it does not match Opus 4 in most domains, offers an optimal combination of capacity and practicality»They explain from the developer.
While the OPUS 4 model is designed for highly professional cases, Sonnet 4 «provides avant -garde performance to everyday cases as an instant update of Sonnet 3.7 ».
Other novelties presented by Anthropic
The IA developer has taken advantage of the launch of her new models to present Other improvements:
Claude Code is now available for the general public: Due to the positive feedback obtained after his previous view of research, Anthropic has expanded the collaboration between developers and Claude. Now, Claude Code admits background tasks using Github Actions and native integrations with vs Code and Jetbrains, showing editions directly in your files to offer a fluid couple programming.
New API capabilities: Four new capacities have been presented at the ANTI Anthropic with which developers and developers can create more powerful agents. These are, the code execution tool, the file API, the MCP connector and the ability to store the indications for up to an hour.
Photo: Anthropic
![5 challenges in digital advertising in an era in which the AI corroles to Google [M4C Sessions]](https://www.logitechgamingsoftware.co/wp-content/uploads/2025/06/m4c-session-channable-5-junio-336x220.jpg)
