Gemini
Gemini is Google DeepMind's family of multimodal models · announced in December 2023 · the models behind the Gemini app and Google's AI products.
Text reviewed October 5, 2026
Gemini is Google DeepMind's family of multimodal models · announced in December 2023 · the models behind the Gemini app and Google's AI products.
Basic
Google DeepMind announced Gemini in December 2023 as a natively multimodal family that handles text, images, audio and video. Each generation ships in tiers: Pro for the hardest tasks, Flash for speed and cost, and lighter variants for high volume. Gemini 1.5 (2024) introduced context windows of one million tokens.
Deep
Gemini models are trained on multimodal data from the start rather than adding vision to a text model afterwards. Google serves them through the Gemini API (Google AI Studio), Vertex AI on Google Cloud and its own products, on its TPU infrastructure. Thinking variants reason before answering, like other labs' reasoning models.
Expert
Google does not disclose parameter counts for Gemini models. Long context is a design focus of the family, but quality at very long lengths still varies by task, so test with your own documents. Google also publishes open-weight models under the separate Gemma name.
Google keeps shipping new Gemini models; the live block on this page lists the newest ones tracked on BenchGecko with their release dates, prices and BenchGecko scores, so this definition never has to name a "latest" model.
Depending on why you're here
- ·Google's AI model family
- ·Understands text, pictures, sound and video
- ·Powers the Gemini app
- ·Flash tiers for cost and speed, Pro for hard tasks
- ·Batch variants are cheaper for work that can wait
- ·See /family/gemini for every tracked model
- ·Runs on Google TPUs, which lowers dependence on NVIDIA
- ·Dated Google figures are on /economy/company/google
- ·Natively multimodal training
- ·Long context is a family focus since Gemini 1.5
- ·Gemma is the separate open-weight line