Gemini 2.5 Deep Think

Published

Aug 1, 2025

Gemini 2.5 Deep Think is an AI model developed by Google and Google DeepMind (United States), first published in August 2025. It works in the language, multimodal, vision and 3 more domain, on tasks such as language modeling/generation, mathematical reasoning, code generation and 6 more.

Epoch AI has no training-compute estimate for this model.

Access: Hosted access (no API). Its weights are not openly released.

Full record

Organization: Google, Google DeepMind
Country of organization: United States
Domain: Language, Multimodal, Vision, Video, Audio, Mathematics
Task: Language modeling/generation, Mathematical reasoning, Code generation, Visual question answering, Question answering, Visual puzzles, Video description, Speech recognition (ASR), Speech-to-text
Model accessibility: Hosted access (no API)
Open weights: No