China’s new artificial intelligence was introduced with advanced capabilities

The Chinese startup “Gipo” on Monday introduced the open-ended model of the GIM-4.3. A model specifically designed for “smart factor” applications. The move comes as Gippo is joining Chinese domestic companies that have released similar models in a growing competition.

Gigpo, a China’s “artificial intelligence tigers”, sponsored by the local government, came to the attention that the US company Open Open in June said the Chinese startup had made significant progress in obtaining government contracts in several areas.

The release of the new Gipo model adds to the growing trend of Chinese models. According to a report from the state -owned news agency Shinhwa, by July, China has released a total of 5 large language models, making it the world’s leading world among the world’s top 5 models.

Capabilities and architecture

The “Game-4.3” model takes a step forward; That is, the essential capabilities of the operating -oriented, such as reasoning, planning, and implementation of action, are directly embedded in the architecture of the model. This approach allows the following:

• Multi -stage analysis of duties and planning
• Using tools and communication with external programming interfaces (API)
• Displaying complex data and managing workflows
• Inherent Support of Reasoning and Perception Cycles – Action

These capabilities pave the way for implementation of operating programs from beginning to end; Programs that were previously implemented only with simple, pre -coded frameworks or closed interfaces.

Permit and pricing access through API

The models of “GM-1-2.3” and “GM-1-1-1” have been published under Apache 2.0; Free and appropriate license for commercial applications.

This license allows developers and organizations to freely and freely distribute these models for research or commercial purposes, editing, personal hosting, optimization and rehabilitation.

For those users who do not want to upload the code or weight of the model themselves, Z.AI cloud services provide access to these models via the API at the following prices:

• GIM 1.1:
  $ 1.2 for each million input token
  $ 1.5 per million token outlet token
• GIM 1.1.1:
  $ 1.5 per million input token
  $ 1.5 per million outlet token

A report from the CNBC news channel reported that the Z.AI will receive only $ 1.5 for each million input to one million and $ 1.5 per million signs for each million. The figure also consistent with a Chinese chart published in the company’s official API documentation.

However, these low prices appear to be applied only to the ceiling of 1.5 input token and 2 output token to each prompt.
(Reminder: Tokens are numerical units that the large language model uses to display semantic concepts and vocabulary components. Each token may represent a complete word or part of a word.)

In fact, the Chinese chart released by the company provides a much more accurate pricing structure for both models, which are based on the input and output packages of the signs.

In short, the launch of the GM-4.3 model to the organizational teams provides a powerful and reliable base model that can be customized, customized and expanded independently; Without dedicated programming or restrictive pricing structures.

This model is an attractive option for teams trying to balance innovation, performance and operational restrictions.