Google DeepMind.

In plain English

Google's strongest current model.

Good for

Complex coding, scientific research, cybersecurity defence and long-running agents.

Things to be aware of

It costs more and runs more slowly than lighter models. Keep it for work where the extra ability changes the result.

In plain English

Google's default model for complex business work.

Good for

Coding, legal and financial analysis, documents with images and charts, and investigation.

Things to be aware of

Legal and financial work still needs qualified review. The model doesn't take professional responsibility for its answer.

In plain English

Google's fast model for work at scale.

Good for

Processing images, audio and text together, software workflows, and agents that use Google tools.

Things to be aware of

Watch what its tools can reach. A cheap model can still create a big security problem.

In plain English

Google’s current family of open-weight models. Gemma can run on company infrastructure, personal computers, and some mobile devices, rather than only through Google’s online services.

Good for

Private or offline AI, custom applications, smaller devices, and organisations that want more control over where a model runs.

Things to be aware of

Gemma 4 comes in several sizes, and they do not perform equally. The smaller versions need less computing power but give up capability. If you run it yourself, updates, security, monitoring, and support become your responsibility.

In plain English

Google’s current model for generating and editing images. You describe a picture or upload an existing one, then use ordinary language to create or change it.

Good for

Product images, visual concepts, adverts, image editing, diagrams, and social content. It is particularly useful when several rounds of changes need to keep the same subject or product.

Things to be aware of

It can still alter faces, products, positioning, or small details that you asked it to preserve. Small text and long paragraphs can come out badly. Check consent, copyright, brand rules, and whether the final image could mislead somebody.

In plain English

A scientific AI model that predicts the three-dimensional structures of proteins and other biological molecules. It helps researchers investigate how molecules may fit together and interact.

Good for

Structural biology, early-stage drug research, studying proteins, and generating scientific hypotheses that can be tested in the laboratory.

Things to be aware of

AlphaFold produces predictions. It does not prove that a molecule behaves that way inside a living system. It can struggle with chemical stereochemistry, molecular movement, and some unusual structures. Scientists still need confidence scores, expert judgement, and experimental evidence.