Google Launches Gemini 1.5 API Models with Enhanced Performance and Lower Costs

Gemini 1.5 Pro and Flash: Google’s Latest API Innovations for Developers

Google has announced the stable release of its Gemini 1.5 application programming interface (API) models for developers, introducing significant performance enhancements and cost reductions in app production. The stable versions, Gemini 1.5 Pro (gemini-1.5-pro-002) and Gemini 1.5 Flash (gemini-1.5-flash-002), were unveiled on September 24 and promise to deliver improved functionality compared to the earlier 001 models.

Significant Improvements in Performance

The newly launched Gemini 1.5 models demonstrate substantial advancements in various areas, including code generation, mathematical accuracy, reasoning, and video analysis. Google reported that these production-ready models are engineered to lower financial barriers for developers, with the Gemini 1.5 Pro model seeing a price reduction of over 50% compared to previous versions.

According to Google’s release notes, both the Gemini 1.5 Pro and Flash models exhibit enhanced factual accuracy and reduced instances of model hallucinations. They also feature improved capabilities in instruction following, multilingual understanding (covering 102 languages), SQL generation, and audio and document comprehension. Additionally, Google has shortened the summarization lengths for both models, providing chat-based product developers with options to enhance the API’s conversational abilities.

Cost Reductions and Enhanced Rate Limits

Starting from October 1, Google will implement significant price reductions for the Gemini 1.5 Pro API. For prompts containing less than 128,000 tokens, prices will decrease by 64% for input tokens, 52% for output tokens, and 64% for incremental cached tokens. This pricing strategy aims to incentivize more developers to utilize the Gemini API for their projects.

In addition, Google announced an increase in the paid tier rate limits: the rate limit for Gemini 1.5 Flash will rise to 2,000 requests per minute (RPM), while Gemini 1.5 Pro will be elevated to 1,000 RPM, up from the previous limits of 1,000 and 360 RPM, respectively.

Experimental Version of Gemini 1.5 Flash

Alongside the stable releases, Google has introduced an experimental version of the Gemini 1.5 Flash API, known as Gemini 1.5 Flash-8B. This smaller model is designed with lower benchmark numbers but includes significant performance increases across both text and multimodal use cases.

All versions of the Gemini API are currently available at Google AI Studio and through the Gemini API.

OpenAI’s Competitive Moves

In a related development, Google’s primary competitor in the artificial intelligence sector, OpenAI, has begun rolling out its new “Advanced Voice” feature to select ChatGPT users. This feature facilitates faster and more intuitive communication with AI, mirroring human-like interactions. OpenAI also unveiled five new voice options—Arbor, Maple, SXol, Spruce, and Vale—to complement its existing voice choices, which include Breeze, Juniper, Cove, and Ember.

FAQ: Google Gemini 1.5 API Release