The artificial intelligence (AI) sector is abuzz with the simultaneous release of new models by industry giants OpenAI, Google, and the French startup Mistral. The launch of these frontier AI models within 12 hours of each other has sparked anticipation for a summer of intensified activity in the industry.
The developments come in the lead-up to the highly anticipated release of the next major version of GPT, the system that powers OpenAI’s popular chatbot Chat-GPT. The flurry of releases commenced with Nick Clegg’s confirmation of Meta’s AI model, Llama, being set for publication in the coming weeks during an event in London.
Shortly after Clegg’s announcement, Google unveiled Gemini Pro 1.5, its most advanced large language model, with a free tier limited to 50 requests a day. Following closely, OpenAI launched its own frontier model, the final version of GPT-4 Turbo. Both GPT-4 Turbo and Gemini Pro 1.5 are ‘multimodal’ systems, capable of processing image inputs, with Gemini also accommodating audio and video inputs.
In the early hours of the morning in France, Mistral, a startup founded by former Meta AI team members, released its frontier model, Mixtral 8x22B, through a simple download link to a 281GB file. Embracing an ‘open source’ approach, Mistral makes its AI systems freely available for download and further development.
While this open approach has been criticized for its potential risks, including the inability to intervene in case of harmful use or address vulnerabilities and biases, proponents argue that it ultimately leads to better outcomes than closed systems.
