Gemini 1.5 Pro’s headline feature was context length. Where most models read a few hundred pages at most, it could take in an hour of video or a whole codebase in one request.
AI model data sheet: Gemini 1.5 Pro
| Developer | Google DeepMind |
|---|---|
| Released | February 15, 2024 |
| Type | Multimodal mixture-of-experts model |
| Context window | Up to 1 million tokens (2 million later) |
| Access | Gemini app, Google AI Studio and Vertex AI |
| Licence / weights | Closed |
| Best at | Very long documents, video and large codebases in one prompt |
| Why it matters | Made million-token context a real product feature |
Quick verdict: Pick it when the input is huge. Check cost per long prompt before scaling.
That made it useful for research, legal review and summarising long recordings.