Gemini 1.5
AI Model
Multimodal
Historical
The million-token context breakthrough.
You're looking at an AI model record: who built it, what it can do, and the hardware and data behind it.
Gemini 1.5 Pro (February 2024) jumped context length to one million tokens — hours of video, whole codebases in one prompt — via MoE architecture and context research that forced every competitor to follow. Trained on TPU v4/v5 fleets.
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Developed by
Trained on
Succeeded by
Sources
Record reviewed as of 2026-01-15 — see Methodology.