Skip to content
Rush Commerce
AI & Automation3 min read

Gemini 4 Neon: $3/$15, near Carbon scores for less

Google's Gemini 4 Neon costs $3/$15 per million tokens with a 1M context and scores 61 on the AA Index. Carbon's promo price ends Nov 30. Rerun your evals now.

Google shipped Gemini 4 Neon on November 18, its fourth Gemini 4 model in seven weeks. Neon sits between Hydrogen and Carbon: $3 per million input tokens, $15 per million output, a 1 million token context, and benchmark scores at or near Carbon's. Google's launch post pitches it as the balance of depth and cost. With Carbon's intro price ending November 30, Neon is the model to test this week.

What actually happened

From Google's announcement:

  • Price: $3 input, $15 output, $0.15 cached input (a 95% discount). No batch or long-context tiers were listed.
  • Context: 1 million tokens, with four thinking levels. High is the default.
  • Availability: paid tier of the Gemini API in Google AI Studio, Gemini for business, Antigravity, and the Gemini app for AI Pro and Ultra subscribers.
  • Role in the family: Google says Carbon handles the hardest long-horizon work, Hydrogen handles fast everyday execution, and Neon "balances depth and efficiency."

Artificial Analysis scored Neon at 61 on its Intelligence Index at its xhigh setting, one point ahead of Carbon and third overall. On its coding agent index, Neon in Antigravity cost $6.86 per task, against $12.61 for Claude Opus 5.5 and $15.47 for Claude Fable 5.2. Its cost advantage comes from price, not efficiency: Neon used about 15% more output tokens per task than Carbon.

9to5Google reports that Carbon and Argon cost $2/$10 until November 30, then $4/$20, and that Neon replaces Argon in the Gemini app. Google's own post does not mention any API deprecation, so don't assume one.

Why it matters for your business

The Gemini 4 price sheet changes in 12 days. If you built on Carbon or Argon at $2/$10, your bill doubles on December 1. Neon at $3/$15 is 25% below that list price and scored about the same on independent tests.

Benchmarks are not your workload. Our order:

  • Run your own eval set on Neon at high, Carbon at high, and your current model. Use 30–50 real tasks from your logs.
  • Measure cost per finished task, not price per token. Neon uses more tokens than Carbon on some tasks.
  • Route by job. Hydrogen for volume, Neon for the middle, Carbon or another frontier model only for the tasks that fail on Neon.
  • Pin model versions in config so a picker change in an app never changes your production output.

Key takeaways

  • Gemini 4 Neon costs $3/$15 per million tokens, with $0.15 cached input and a 1M context
  • Artificial Analysis scores it 61, one point above Carbon and third overall
  • Carbon and Argon go from $2/$10 to $4/$20 after November 30, per 9to5Google
  • Neon's lower cost comes from token price; it uses more tokens per task than Carbon
  • Test on your own tasks and route by job before December 1

Not sure which Gemini model your app should run? We build eval harnesses and model routers that measure cost per finished task, so a price change is a config edit, not a rewrite. See how we build it or run the numbers on your AI spend.

Sources: Google, Artificial Analysis, 9to5Google.

  • #gemini-4-neon
  • #google-gemini
  • #llm-pricing
  • #model-routing
  • #ai-coding
TR

Tommy Rush — Founder, Rush Commerce

Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More

Get The Rush Report weekly — one email, zero fluff.