Google renames Bard to Gemini and launches Gemini Advanced with Ultra 1.0
The $19.99-a-month Google One AI Premium tier gave access to Ultra 1.0, which Google said was the first model to outperform human experts on MMLU.
- Models & capabilities
- Notable
Google chief executive Sundar Pichai announced that Bard, the chatbot Google had launched a year earlier as a lightweight assistant built on LaMDA, would be retired as a brand name and folded into “Gemini” — the name of the multimodal model family Google had introduced in December 2023. The consumer product, available in 40 languages across web and mobile, became simply Gemini.
Alongside the rename, Google launched Gemini Advanced, a paid tier giving access to Ultra 1.0, the largest model in the Gemini family, through a new subscription called Google One AI Premium priced at $19.99 a month with a free two-month trial, available in English across roughly 150 countries. Pichai’s announcement cited Ultra 1.0’s MMLU score as the first to exceed the performance of human experts on that benchmark, framing the launch as evidence the model had reached genuine frontier capability rather than parity with existing systems.
The rebrand consolidated what had been a confusing set of overlapping names — Bard for the consumer chatbot, Gemini for the underlying model family, and Duet AI for Google’s enterprise assistant products, which was rebranded to Gemini for Workspace the same week — into a single product identity spanning free and paid consumer tiers, developer APIs, and enterprise tools. It followed the pattern set by OpenAI’s ChatGPT/GPT split and Microsoft’s consolidation of its assistants under the Copilot name, and marked the point at which “Gemini” replaced “Bard” as Google’s public answer to ChatGPT.
The Ultra 1.0 tier proved short-lived as a standalone flagship: within the week, Google shipped Gemini 1.5 Pro, a smaller mixture-of-experts model that matched much of Ultra’s performance at lower compute cost and a far longer context window, and which became the more widely used model in practice.