No longer supported

This model is deprecated (released February 2024) and is no longer actively maintained or recommended by Google. Consider using their current flagship model instead.

deprecatedfrontiermultimodal Proprietary

Gemini 1.0 Ultra

by Google· Released February 2024· Cutoff Early 2023

Gemini 1.0 Ultra is Google's largest and most capable model in the Gemini 1.0 family, designed for highly complex tasks. It excels in reasoning, multimodal understanding, and coding, but has been superseded by Gemini 1.5 Pro and Gemini 1.5 Flash. It is no longer available via the API as of early 2025.

Official Site API Docs 📄 Research Paper

Input cost

$5.00 per 1M tokens

Output cost

$15.00 per 1M tokens

Context window

32K tokens

Max output

8192 tokens

Modalities

textimageaudiovideo

License

proprietary

Capabilities

Multimodal understanding (text, image, audio, video)ReasoningCode generationFunction callingJSON modeStreaming

Best For

Complex reasoning and multimodal tasks requiring the highest capability in the Gemini 1.0 series.

Strengths

State-of-the-art reasoning and problem-solving
Strong multimodal performance across text, image, audio, and video
Excellent coding and math capabilities

Limitations

Deprecated and no longer accessible via API
Smaller context window (32K) compared to later models
Higher latency and cost than newer Gemini models

Use Cases

Advanced research and analysis

Complex code generation and debugging

Multimodal content understanding (e.g., video analysis)

Scientific reasoning and problem-solving

High-stakes decision support

Educational tutoring for advanced topics

Creative content generation with multimodal inputs

Improvements Over Previous Model

First model in the Gemini family, establishing new benchmarks in multimodal understanding
Introduced native multimodal capabilities across text, image, audio, and video
Set state-of-the-art results on 32 of 32 academic benchmarks at launch
Outperformed GPT-4 on several reasoning and coding tasks
First Google model to achieve 'frontier' tier status

Back to all models

Improvements Over Previous Model

First model in the Gemini family, establishing new benchmarks in multimodal understanding

Introduced native multimodal capabilities across text, image, audio, and video

Set state-of-the-art results on 32 of 32 academic benchmarks at launch

Outperformed GPT-4 on several reasoning and coding tasks

First Google model to achieve 'frontier' tier status