I noted Gemini 4 Neon as Google’s middle-tier model, positioned between fast Hydrogen and flagship Carbon. Its reported benchmark and cost claims make it a useful comparison point for gemini 4 carbon and gemini 4 hydrogen, though the post says the headline figures are Google’s own.
Gemini 4 Neon launched on 18 November.
Gemini 4 Neon is positioned between the fast Hydrogen and the flagship Carbon.
The post reports that Neon scores 61 on the Artificial Analysis Intelligence Index.
The post reports that Neon ranks third overall on the Artificial Analysis Intelligence Index.
The post reports that Neon ranks first on the Coding Agent Index at 58.6.
The post reports a score of 65.2% on Terminal-Bench 4.1.
Google says Neon costs 32% less per completed task than Carbon at its default setting.
The post says Neon leads on Zapier's AutomationBench.
The post says Neon leads on a long-horizon vending-machine business simulation.
Neon is priced at $3 per million input tokens.
Neon has a one-million-token window.
The post says every headline figure is Google's own.
The post says the full safety report is still to come.
I think a mid-tier model beating the flagships on coding is a pattern worth watching.
I noted the post’s question: Is the next race about the biggest model, or the one that finishes the job cheapest?