D. Jassim Haji
The company “Mita” has announced the launch of the Llama API API application, designed to help developers to integrate the “Lama” artificial intelligence models in their products more easily. This interface was presented during the first conference of artificial intelligence developers organized by “Mita”, and it is currently available for a selection of users, with a plan to expand the scope of availability in the coming months.
Parallel to the launch of the interface, Mita launched an independent application of an artificial intelligence assistant, and confirmed her intention to test a subscription system paid for artificial intelligence talks later this year. The company also highlighted the improvements that have occurred on the latest “Lama” models, focusing on efficiency and reducing costs.
“Mita” also confirmed that the developers will keep full control of the dedicated models that they create, with the flexibility in spreading them outside the “dead” platforms.
The launch of the “Lama” interface comes within the framework of the “Mita” seeking to preserve its leadership in the open source models market, which is witnessing sharp competition. Although Lama’s models have been loaded more than a billion times so far, according to Mita, competitors such as Deepseek and QWEN (QWEN) affiliated to the company “Ali Baba” are a challenge to “Mita” aspirations in building a wide system based on “Lama”.
The Lama interface provides tools to customize models and evaluate their performance, starting with the Lama 3.3 B model. Users can generate, train training data, then use the evaluation tools set at the Lama interface to test the specific model quality.
“Mita” confirmed that it will not use the data of the “Lama” interface customer to train its own models, and that the models that are created through this interface can be transferred to other hosting platforms.
Mita provides Lama models for developers almost free, a strategy for CEO Mark Zuckerberg has indicated that she will bear fruit by stimulating innovation, reducing dependence on potential competitors, and increasing interaction on the main social networks of the company.
– “Meta – Serbras”: This partnership allows developers working on the “Llama 4” model of “Serbras” via the API interface (API). This high speed is a gateway to a completely new generation of applications that it was impossible to build using other technologies. These applications include: immediate, low -cumin audio conversations, the generation of interactive codes, and immediate multi -step inference, as well as smart agents in real time – all depend on a series of large linguistic models (LLM), which can be implemented in seconds instead of minutes.
