Google has rolled out three experimental AI models as part of its Gemini 1.5 series. The updated collection aims to improve both performance and user experience. The fresh additions include the Gemini 1.5 Flash-8B, a compact version of the existing Flash model; the Gemini 1.5 Pro, which excels in coding and handling intricate prompts; and an upgraded variant of the Gemini 1.5 Flash.
Gemini 1.5 Flash-8B Details
The Gemini 1.5 Flash-8B, equipped with 8 billion parameters, targets high-volume multimodal applications as well as tasks requiring extensive summarization. The streamlined version seeks to expand on the established success of its predecessor. It’s currently available for free to developers via Google AI Studio and the Gemini API under the identifier “gemini-1.5-flash-8b-exp-0827.”
Key improvements are evident in the new Gemini 1.5 Flash and Gemini 1.5 Pro models. Logan Kilpatrcik, head of Google AI Studio, introduced the new models on X. The latest Flash model shows noticeable enhancements across several benchmarks, whereas the Pro model excels in coding tasks and complex prompt handling. These models are accessible through the Gemini API and Google AI Studio under “gemini-1.5-pro-exp-0827” and “gemini-1.5-flash-exp-0827,” respectively. Google anticipates making these models available for broader production use in the coming weeks.
Today, we are rolling out three experimental models:
– A new smaller variant, Gemini 1.5 Flash-8B
– A stronger Gemini 1.5 Pro model (better on coding & complex prompts)
– A significantly improved Gemini 1.5 Flash modelTry them on https://t.co/fBrh6UGKz7, details in 🧵
— Logan Kilpatrick (@OfficialLoganK) August 27, 2024
Community Feedback and Future Developments
The new Gemini 1.5 models have undergone extensive testing in Chatbot Arena, garnering more than 20,000 community votes. Gemini-1.5-Flash (0827) notably climbed from twenty-third to sixth place. Moreover, the Flash-8B model surpasses Google’s previous Gemma-2-9b performance and is on par with Llama-3-70b. The Pro (0827) model now occupies the fourth position in coding and has shown marked improvement in mathematical tasks.
Developers can test these models on Google AI Studio. Through these updates, Google is advancing its AI capabilities and is keen to receive feedback from its user community. Earlier this month, Google announced Gemini Live, a voice-supporting AI assistant version of the model. Announced during the Pixel 9 event, this voice chat mode is now available for Gemini Advanced subscribers, enabling real-time voice conversations similar to ChatGPT’s capabilities.
Last Updated on November 7, 2024 3:05 pm CET
