Longterm Wiki

Gemini 1.0 Ultra

Google DeepMind

Gemini 1.0 Ultra launched February 8, 2024 via Gemini Advanced. First model to exceed human expert performance on MMLU (90.0%). Built natively multimodal, processing text, images, audio, and video in a single model. Powered the Gemini Advanced subscription tier.

Developer
Google DeepMind
Released
2024-02-08
Context Window
128K tokens

Benchmarks2

BenchmarkScore
MMLU90%
HumanEval74.4%

Gemini Family5

ModelTierReleasedInput $/MTok
Gemini 2.5 Flashflash2025-05-20$0.15
Gemini 2.5 Propro2025-03-25$1.25
Gemini 2.0 Flashflash2024-12-11$0.1
Gemini 1.5 Flashflash2024-05-24$0.075
Gemini 1.5 Propro2024-02-15$1.25

Details

Model FamilyGemini
Tierultra
Generation1.0
Release Date2024-02-08
Context Window128K tokens

Capabilities3

tool-usevisionaudio-input

Sources1

Tags

geminigooglemultimodalfrontier-ai