Google launches Gemini 4 Neon, citing lower cost per task and stronger coding results
Access starts today across the Gemini API, Gemini for business, paid Antigravity plans and Pro and Ultra subscriptions in the Gemini app.

Google has launched Gemini 4 Neon, which it positions as a frontier model for long-running software engineering, knowledge work and scientific tasks. On the Artificial Analysis Intelligence Index it places third overall, with a score of 61 at its highest thinking tier, and it ranks second on the Vals Index. Pricing is set at three dollars per million input tokens, with cached input reduced sharply, and fifteen dollars per million output tokens, within a one-million-token context window. Access begins today through the Gemini API, Gemini for business, paid Antigravity plans and Pro or Ultra subscriptions in the Gemini app. For developers, the notable claims are the lower cost per task at its default setting and its stronger results on coding and cyber-defense benchmarks.
Why it matters
For people who follow the AI model race, Neon is a clear step in Google's long-running push for frontier models aimed at extended software, knowledge and scientific work. Its ranking near the top of established indexes, combined with a lower cost per task at its default setting than the earlier Carbon model, suggests Google is competing on price-performance as well as raw capability. Readers should weigh these results as vendor-reported benchmark positions rather than independent proof of how the model performs in everyday use.
Matters or hype?
Does this story matter, or is it hype? Vote before you see what everyone else said.
Sources · 1
- Google Blog ↗ Primary source