OPT vs. Gemini 4

Large language models, which often demand significant computational power and prolonged training periods, have shown remarkable abilities in performing zero- and few-shot learning tasks. The substantial resources required for their creation make it quite difficult for many researchers to replicate these models. Moreover, access to the limited number of models available through APIs is restricted, as users are unable to acquire the full model weights, which hinders academic research. To address these issues, we present Open Pre-trained Transformers (OPT), a series of decoder-only pre-trained transformers that vary in size from 125 million to 175 billion parameters, which we aim to share fully and responsibly with interested researchers. Our research reveals that OPT-175B achieves performance levels comparable to GPT-3, while consuming only one-seventh of the carbon emissions needed for GPT-3's training process. In addition to this, we plan to offer a comprehensive logbook detailing the infrastructural challenges we faced during the project, along with code to aid experimentation with all released models, ensuring that scholars have the necessary resources to further investigate this technology. This initiative not only democratizes access to advanced models but also encourages sustainable practices in the field of artificial intelligence.

Gemini 4 is Google’s next major Gemini model family, currently confirmed as being in pre-training rather than publicly released. The model follows recent Gemini releases such as Gemini 3.6 Flash and Gemini 3.5 Flash-Lite, which focused on efficiency, agentic workflows, coding, multimodal tasks, and lower-cost production AI. Google has described Gemini 4 as its most ambitious pre-training run yet, suggesting that it is intended to push the company’s frontier AI capabilities forward. As of now, Gemini 4 does not have an official public launch, model card, pricing page, API documentation, benchmark suite, or confirmed availability timeline. Because of that, any specific claims about context length, model sizes, exact capabilities, pricing, or release channels should be treated as unconfirmed until Google publishes official details. Based on Google’s current Gemini direction, Gemini 4 is expected to improve areas such as advanced reasoning, software engineering, multimodal understanding, AI agents, knowledge work, and enterprise AI workflows. It may eventually power products across the Gemini app, Gemini API, Google AI Studio, Gemini Enterprise, Google Cloud, and other Google services. The model is also likely to be important for developers building production AI systems that need reliable reasoning, tool use, speed, and scalable deployment options. For enterprises, Gemini 4 could become a foundation for AI assistants, workflow automation, document analysis, code generation, customer support, and internal knowledge tools. For now, the best way to describe Gemini 4 is as Google’s confirmed next-generation Gemini model effort, not as a generally available product. By extending the Gemini roadmap beyond the 3.x series, Gemini 4 represents Google’s next step toward more powerful, multimodal, and agentic AI systems.