Microsoft announced the launch of two new artificial intelligence models dedicated to programming tasks, "MAI Code 2" and "MAI Code 2 Flash", with the aim of helping organizations develop software and carry out everyday programming tasks with greater efficiency and at lower cost.
In a statement published on its official website yesterday, Tuesday, coinciding with the "Microsoft Ignite 2026" conference, the company said the two models were designed to balance performance quality, execution speed and cost. It explained that the first model targets advanced and complex programming tasks, while the "Flash" version focuses on repetitive, fast, high-volume tasks.
Capabilities of the two models
Microsoft explained that the "MAI Code 2" model has capabilities that include handling complex programming errors, making changes that span multiple files, analyzing code repositories, and developing user interfaces, as well as turning code into products ready for release, while taking into account the maintenance and review standards adopted by development teams.
By contrast, the "MAI Code 2 Flash" model was designed to carry out repetitive everyday tasks quickly, such as explaining code, making specific modifications and fixing routine errors, which helps speed up development processes and reduce operating costs.
Test results according to the company
According to the company, the "MAI Code 2" model achieved about 98 percent of the performance of the "GPT-6.1 Sol" model at 40 percent of its cost, and scored 96 percent of the performance of the "Claude Opus 5.5" model at 5 percent of its cost, based on the average results across four benchmark tests specialized in everyday enterprise programming tasks.
The company added that the model has capabilities in designing dashboards and visual representations and developing application interfaces, noting that the results of blind evaluations of application interface design showed reviewers preferred its outputs over those of the "GPT-6 Sol" model by 64 percent.
As for the "MAI Code 2 Flash" model, according to Microsoft, it delivers performance close to that of the "GPT-6 Luna" and "Claude Haiku 5.5" models, at a cost of 65 percent and 20 percent of the cost of each, respectively, based on three benchmark tests that measure the models' efficiency in everyday and repetitive programming tasks.
Context of the launch
The launch of the two models comes amid growing competition among technology companies to develop AI tools specialized in programming and to provide solutions that help organizations speed up software development and improve operational efficiency.
Source: Syrian Arab News Agency (SANA)

Comments (0)