Dive Brief:
- Google is giving enterprises access to its Gemini Pro LLM via Google Cloud’s Vertex AI platform, the company announced in a blog post Wednesday. Developers can also use the model via the Gemini API in Google AI Studio.
- The Google AI Studio is a free, web-based developer tool that allows 60 requests per minute. Users can transition to Vertex AI from Google AI Studio for additional security, privacy, data governance and compliance features, according to the company.
- Gemini was released last week in three sizes: Ultra, Pro and Nano. The Pro size also powers Google Bard, and can process text. Google created a dedicated Gemini Pro Vision model, which is currently available, to process imagery inputs.
Dive Insight:
Google’s highly anticipated Gemini family of LLMs is the latest attempt from big tech providers to seek enterprise buy-in for generative AI. Microsoft has laid out plans to equip tools with the latest OpenAI model, GPT-4 Turbo, and Amazon has started to make its Titan models generally available as well.
Users accessing Gemini Pro and Pro Vision through Vertex AI will do so at no additional cost until general availability early next year when the pricing structure will change to charge per 1,000 characters and input image across Google AI studio and Vertex AI. The future pricing structure differs from other vendors, which typically track token usage.
Google is planning to launch Gemini Ultra, the largest of the three sized models, early next year after further fine-tuning, safety testing and gathering feedback from its partners, according to the blog post.
Enterprises’ options have expanded as more foundational models hit the market, but big tech providers still have ample room to improve when it comes to transparency.
Regulatory noise is intensifying, too. The clock is ticking for providers to share more about how their technology products work. Part of the provisional agreement tied to the European Union’s AI Act released Friday required developers of general purpose AI models to draw up technical documentation and disseminate detailed summaries about training content.
Noncompliant developers face fines ranging from 7 million to 35 million euros, depending on the infringement and size of the company, according to the initial agreement.
As organizations ramp up generative AI experimentation, experts say now is the time for enterprises to think critically about selected vendors, responsible frameworks and governance.
