FLASH
Shkruar nga Pamfleti
Thinking Machines Lab, the company founded by former OpenAI CTO Mira Murati, has unveiled its first general-purpose artificial intelligence model, based on several technologies developed by Chinese companies.
According to the Financial Times, the new model called Inkling is built on the DeepSeek-V3 architecture and has been improved after the initial training phase with data generated by the Kimi K2.5 model from the Chinese company Moonshot AI.
Inkling's launch comes as Chinese AI labs are increasing their influence in the global tech race. According to OpenRouter, Chinese AI models have surpassed American models in overall usage this year.
Unlike more advanced models from companies like OpenAI and Anthropic, Inkling will be an "open-source" model, meaning users will be able to install it on their own servers and customize it to their specific needs.
"Today we are advancing our mission by releasing a model that we have trained from scratch, with all the available weights, so that people can make it their own," the company stated.
Thinking Machines Lab has reached a $12 billion valuation just a year after securing $2 billion in seed funding. Its investors include big names in the tech industry, including Andreessen Horowitz, Nvidia, AMD and Jane Street Fund.
Mira Murati became one of the most prominent figures in the AI industry during her time leading the development of products like ChatGPT, DALL-E, and OpenAI's voice capabilities. She left the company in September 2024 and founded Thinking Machines Lab in February of the following year.
However, despite the great attention the new company has received, it has also faced some challenges. This year, several senior executives left to join companies like Meta and OpenAI, while discussions have also arisen about Murat's leadership style.
In addition to Inkling, the company is also developing the Tinker platform, which aims to enable businesses to customize large language models according to their requirements.
Lini një Përgjigje