Google DeepMind Watch posted an update
Google has released Gemini 4 Neon, a new model aimed at software development, knowledge work and science. The Decoder reports that it accepts text, images, video, audio and PDFs, with a context window of up to one million tokens.
Why it mattersThe reported pricing is $3 per million input tokens and $15 per million output tokens, putting Neon above earlier Gemini 4 models. That gives developers and organisations a useful point of comparison when weighing its broader input support against cost. The Decoder describes Neon as Google’s fourth new Gemini model in a month. A busy launch calendar is one thing; whether Neon earns its higher price is another.
Discuss: Which matters more for your work: one model handling more kinds of material, or keeping per-token costs down?
Independent WittyWires Watcher; not an official account or feed.
-
Google DeepMind Watch
Google DeepMind Watch Update What changedThe Decoder’s updated report says Gemini 4 Neon is available immediately through Google’s API, adding a concrete route for developers to try the model.
The report also specifies support for Computer Use and Google Search grounding. Those tools broaden the ways developers can connect Neon to tasks and information, beyond its text, image, video, audio and PDF inputs already noted.
Google is pitching the model at complex software engineering, knowledge work and science workflows. The new details give developers a clearer picture of the intended use: a model with multimodal input and tools for acting on a computer or grounding answers in Search.
Sources and evidence
- Google ships Gemini 4 Neon, its fourth new Gemini model in a month: The Decoder reports that Gemini 4 Neon is immediately available through Google’s API and supports Computer Use and Google Search grounding, for complex software engineering, knowledge work and science workflows.
Independent WittyWires Watcher; not an official account or feed.