Gemini 4 Neon adds another option for commerce AI work

Google introduced Gemini 4 Neon on 18 November, pitching it for complex coding, knowledge work and long-running agent workflows. For commerce teams, the immediate decision is whether it can finish a…

JJames Bennett1 min read

Google introduced Gemini 4 Neon on 18 November, pitching it for complex coding, knowledge work and long-running agent workflows. For commerce teams, the immediate decision is whether it can finish a useful task reliably at an acceptable total cost.

Google lists API pricing of $3 per million input tokens, $0.15 per million cached input tokens and $15 per million output tokens, with a one-million-token context window. Those are token rates, not a fixed price for completing an audit or building an app.

Where the model is arriving

The launch describes rollout through the Gemini API in Google AI Studio, Gemini for business, and Google Antigravity for paid Pro and Ultra subscribers. It also reaches Pro and Ultra subscribers in the Gemini app. Availability in one surface should not be treated as evidence of identical access or billing in another.

Neon offers four thinking levels. Google says its default high setting delivers a 32% lower cost per task than Carbon on the cited comparison, while xhigh targets harder reasoning. These are benchmark claims from the launch, not savings measured on a merchant’s catalogue or support workload.

A useful commerce evaluation

Our suggested starting point is one repeatable task: reconcile a product feed against a source spreadsheet, investigate a small app bug, or produce a report from authorised business documents. Use a known answer or a human-reviewed sample, and record errors, elapsed time, retries and total spend.

A large context window can hold more evidence, but it does not guarantee that every price, variant or policy will be interpreted correctly. Keep write actions behind review until the workflow has passed your own tests. Google’s model benchmarks do not establish a new Shopify integration, automatic access to store data or guaranteed commercial results.

Google’s launch post provides the announcement graphic and model positioning.

ℹ️
Compare total task cost and correctness at the same thinking level. Published token prices and benchmark rankings do not predict your store’s outcome.
Filed underAI

More from AI

See all →