Gemini 3 Ultra
FrontierMultimodalAPI Available
Benchmark Scores
93.6/100MMLU94.5
HumanEval93
Math92.5
Reasoning94.8
Coding93.2
About Gemini 3 Ultra
Gemini 3 Ultra is Google's most capable model, featuring a massive 2M token context window and native multimodal understanding across text, images, video, and audio. It excels at tasks requiring long-context reasoning.
Strengths
- +2M token context window
- +Native multimodal reasoning
- +Strong at long-document analysis
- +Excellent video understanding
Weaknesses
- -High latency on maximum context
- -Expensive at scale
- -Occasional formatting inconsistencies
Pricing
Per 1M tokens
Input$7.00
Output$21.00
Quick Facts
- Context Window
- 2M
- Parameters
- Undisclosed
- Release Date
- 2026-02-20
- Category
- Frontier
Compare Models
See how Gemini 3 Ultra stacks up against other models