On Tuesday, Google released newer versions of its Gemini 1.5 Pro artificial intelligence (AI) model. They come just months after the Mountain View-based tech giant released the latest version of Gemini, which increased the context window to up to 2 million tokens. Called Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002, the company said these AI models not only offer higher output power and lower costs, but will also enable higher speed limiting for users. Additionally, the filter settings have also been updated to help the AI ​​models better adhere to the instructions.

New Gemini AI models released

In a blog post, the company detailed the Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002 AI models. These models are currently available as experimental model releases and are built on top of Gemini 1.5 Pro which was first announced at Google I/O in May. They are currently available to developers and business users of the company. Developers can access it for free from Google AI Studio and the Gemini API. Businesses can access it through Vertex AI.

The latest Gemini 1.5 Pro and Flash models also outperformed the previous generation Gemini model, according to internal testing conducted by Google. The company claimed the new models saw a seven percent increase in the Massive Multitask Language Understanding Pro (MMLU-Pro) benchmark. Furthermore, the AI ​​models are said to offer roughly a 20 percent improvement in the MATH and HiddenMath benchmarks compared to Gemini 1.5 Pro.

The Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002 AI models will also offer an increased speed limit. Rate caps are daily usage limits for users. With the 1.5 Flash model, users will get 2,000 requests per minute (RPM), and the 1.5 Pro model will offer 1,000 RPM. Google said those limits are being increased to allow developers to build with new versions of Gemini.

Only the rate limit is not upgraded. With the Gemini-1.5-Pro-002 and Flash-002, the company has also increased the output tokens per second, making the models more responsive and faster to generate long blocks of text.

One key upgrade with these AI models is improvements to filters. Google said the new Gemini AI models will adhere to prompts and follow instructions better thanks to these updated filters. Google is also improving its suite of security measures to ensure that AI models don’t create anything harmful. Notably, the default filters will not be applied in the new AI models so that developers can choose the desired configuration.