Google launches a new upgraded version of the Gemini 1.5 model series, Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002 make a shocking debut! The editor of Downcodes learned that this update not only brings a significant improvement in performance, but also a significant price reduction, which can be described as major good news in the field of AI development. The performance improvement is astonishing, and the cost is reduced by more than half, which is undoubtedly very attractive to developers and enterprise users. Let’s take a look at what exciting improvements this update brings.
Google today announced the launch of a new upgraded Gemini model series, including Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002. This update not only greatly improves performance, but also brings surprising price discounts, which will undoubtedly set off a boom in the AI development circle.
First of all, the most eye-catching thing is the significant price reduction. The cost of using the new version of the model has been cut in half, by more than 50%. At the same time, performance has been significantly improved. The processing speed of Gemini1.5Flash is doubled, while that of Gemini1.5Pro is nearly tripled. This means that developers can obtain faster output and shorter latency at a lower cost, greatly improving development efficiency.
In terms of performance, the new Gemini model shows comprehensive improvements. Especially in mathematics, long text processing and visual tasks, the progress is particularly obvious. For example, models can now easily process PDF documents of more than 1,000 pages, answer questions containing tens of thousands of lines of code, and even extract useful information from a one-hour video. On the challenging benchmark test MMLU-Pro, the new model achieves approximately 7% performance improvement. Even more impressive is that in the MATH and HiddenMath benchmarks, the improvement is as high as 20%.
Google has also optimized the model’s response quality. The new version provides more helpful and concise answers while keeping the content safe. In tasks such as summarization, question and answer, and information extraction, the output length is shortened by 5% to 20% compared with before, which not only improves efficiency but also further reduces usage costs.
For enterprise users, Gemini1.5Pro’s long text processing capabilities (up to 2 million words) and multi-modal capabilities open up new application scenarios. Starting from October 1, 2024, the prices of input tokens, output tokens and incremental cache tokens will be reduced by 64%, 52% and 64% respectively, which will undoubtedly greatly reduce the cost of using AI for enterprises.
Google has also made significant adjustments when it comes to usage restrictions. Gemini1.5Flash's paid service rate limit is increased to 2,000 requests per minute, and 1.5Pro is increased to 1,000 requests, which provides developers with greater flexibility.
Security has always been a key focus area for Google. The new version of the model further enhances security while following user instructions. Google has also tweaked the default settings for filters to give developers more autonomy.
In addition, Google also launched an improved version of the Gemini1.5Flash-8B experimental model, which performs well in text and multi-modal application scenarios. This version has been released on Google AI Studio and Gemini API, providing developers with more choices.
For Gemini Advanced users, they will soon be able to experience the Gemini1.5Pro-002 version optimized for chat.
Reference: https://developers.googleblog.com/en/updated-production-ready-gemini-models-reduced-15-pro-pricing-increased-rate-limits-and-more/
All in all, the upgrade of the Gemini 1.5 model series has brought huge benefits to AI developers and enterprises. Lower costs, faster speeds, more powerful performance, and more convenient use experience indicate the future of AI applications. Brighter. The editor of Downcodes looks forward to seeing the wonderful applications of the Gemini 1.5 series in various fields!