Google recently unveiled Gemini 1.5, the latest iteration following the launch of Gemini 1.0 just a few months ago. This upgraded version boasts “significantly improved performance,” according to Google. Gemini 1.5 Pro has some features, one of which is a significantly larger context window. This development will be particularly appealing to users who enjoy using the chatbot function.
The all-new Google Gemini 1.5
The Gemini 1.5 Pro Google tier offering now comes with a standard context window expanded to 128,000 tokens, compared to 32,000 tokens. With this update, it becomes capable of translating, over 700,000 words and processing codebases with more than 30,000 lines of code. It can also handle 11 hours of audio and 1 hour of video. In comparison, GPT 4 Turbo also offers a context window of 128,000 tokens, while Claude 2.1 provides up to 200,000 tokens.
Google plans to make up to one million tokens for production testing by early adopters after successfully testing over ten million tokens. These efforts highlight Google’s commitment to improvement and innovation in Gemini 1.5 and future updates. The advancements in Gemini 1.5 are made possible by the Mixture of Experts (MoE) architecture, which divides models into expert neural networks. This architecture improves efficiency in both training and serving updates.
In terms of performance, Google claims that Gemini 1.5 Pro outperforms its predecessor on around 87% of benchmarks across domains such as text analysis, code evaluation, image recognition, audio processing, and video comprehension. At the moment, it seems that Gemini 1.5 Pros performance is said to be similar, to that of the 1.0 Ultra edition.
Nevertheless, the availability of Gemini 1.5 Pro will be postponed as Google gradually introduces the expanded context window to developers and enterprise customers in a controlled manner. This is a phase, which means there will be a wait before it becomes widely accessible.
Despite the wait, anticipation for Gemini 1.5 Pro remains high, reflecting the enthusiasm among fans for Google’s continuous enhancement of its AI model. The gradual rollout suggests accessibility to the broader audience shortly, promising significant impact once it becomes widely available.
