Anthropic has announced two major updates for its Cloud AI model – how much information they can handle at once in a big leap, and a special offer for the US government that recently matches and expands a step by OpenaiI. The company said it would provide its cloud model available to all three branches of the US government – executive, legislative and judiciary – only $ 1 per year. This proposal will be valid for one year and its purpose is to use cloud more widely in public sector work. OpenaiI revealed just a week after the disclosure, it is discovered that the chatup enterprise will be offered to the entire Federal Executive Branch for the same $ 1 annual fee. Anthropic deal moves forward by targeting every branch of the government, not only the executive branch.

Long memory for cloud sonnet 4

With the government’s push, Anthropic is making a major improvement in its Cloud Sonnet 4AI model. The model’s reference window (the amount of information can process at once) has now been increased to 1 million tokens. It is similar to about 750,000 words or about 75,000 lines are code. To keep it in another way, the model can take more than the entire Lord of Rings Triology in the same prompt.

The new border is five times larger than the first 200,000 -token capacity of Cloud and doubles OpenaiI’s GPT -5 to over 400,000 token limits. Upgrade developers and enterprises will make the cloud more capable, which need to work with large amounts of data at once, such as analyzing large -scale codebase or processing long research documents.

Enterprise customers will be able to use this long reference window through API of anthropic. This feature will also be available through its cloud partners, including vertex AI of Amazon Bedrock and Google Cloud, which is accessible to organizations already working on those platforms.

Focus on the government with strong safety standards

Anthropic will provide two main versions of its AI for customers – cloud for enterprise and cloud for the government. The government-specific offer is designed to support the “Fedemp High” workload, which is the most strict security standard to handle sensitive but unplaced US government data.

This means that federal employees can use clouds for the tasks that include confidential information while living within the official safety requirements. Anthropic says that Cloud can help in various types of government related tasks, in which from accelerating scientific research to improving civil services. For example, the model can be used to abbreviate complex documents, translate information in many languages or aid in technical projects.

The company has also planned to provide technical assistance to the agencies so that they can integrate Cloud in their current system. Due to the partnership with AWS, Google Cloud and Palaantir, agencies can use the cloud through their current safe infrastructure without making major changes in their setups.

Part of a big AI competition

This step is the latest in the ongoing fight between major AI companies to play a big role in government projects. Recently, the US General Services Administration added anthropic, openi and Google Deepmind to its list of AI vendors approved for civilian agencies. Being in this list makes it easier for government bodies to buy their services.

Openai has already proposed his $ 1-year-old proposal to the Executive Branch, and those who monitor the industry are now waiting to see if Google will respond with a similar deal.

Anthropic is already involved in AI work associated with the government. It is one of several companies with Openai, XAI and Google, which has been allowed up to $ 200 million by the US Defense Department to use AI for national security purposes. But the latest announcement of the company shows that it is looking into areas such as defense and scientific research and public health.