Gemini 4 Neon Pricing: and Per Million Tokens, and Where It Sits Against Carbon and Argon
Google made Gemini 4 Neon (model id gemini-4-neon) generally available on the Gemini API on November 18, 2026, priced at .00 per million input tokens and .00 per million output tokens on the paid tier. It has a 1M token context window, is not offered on the free tier, and is aimed at software engineering, knowledge work and science workflows, so teams choosing a Gemini 4 model for coding agents or long document work now have a fourth option with a fixed price and no introductory rate.
Key facts
- November 18, 2026: Gemini 4 Neon listed as generally available in Google's Gemini API changelog.
- Standard paid tier: .00 per 1M input tokens, .00 per 1M output tokens (thinking tokens count as output), return [ .15 per 1M for context caching.
- Batch and Flex: .50 input and .50 output per 1M tokens, caching return [ .075. Priority: .40 input and .00 output, caching return [ .27.
- Context window of 1M tokens; the models page lists max output as 64K / 1M tokens, and the changelog says long decode reaches up to 1M tokens.
- Four thinking levels: low, medium, high (the default) and xhigh.
- Not available on the free tier. Only implicit caching applies; explicit caching is not supported.
What Google released
The changelog entry describes Neon as bringing frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work and science workflows. It accepts text, image, audio, video and PDF input. The same November 18 entry added a second change worth knowing: the Gemini API now takes Microsoft Office and OpenDocument files directly, and lets you slice documents, spreadsheets, video and audio so you pay only for the part you select.
Google's pages do not state how Neon ranks against Gemini 4 Carbon or Argon on quality. The comparison below is on price and published limits only, which is what you can verify today.
How the price compares inside the Gemini 4 family
Gemini 4 Hydrogen, generally available since November 12, costs .50 input and .50 output per million tokens, so Neon is exactly twice Hydrogen's rate. Gemini 4 Carbon and Gemini 4 Argon both run at an introductory .00 and .00 through November 30, 2026, then move to .00 and .00 from December 1. That gives Neon an odd position: for the next ten days it costs more than Carbon and Argon, and from December 1 it costs 25 percent less than both on input and output.
A worked example using the published standard rates, for a month with 10 million input tokens and 2 million output tokens:
- Hydrogen: input plus output, .
- Neon: input plus output, .
- Carbon or Argon from December 1: input plus output, .
- Neon through the Batch tier: plus , .
Output includes thinking tokens, so a request run at the xhigh thinking level will bill more output than the same request at low. Measure your own output token counts per level before you budget. We covered the Argon step-up in detail in Gemini 4 Argon Pricing Doubles on December 1, and Hydrogen against the older Flash line in Gemini 4 Hydrogen Pricing.
Limits that affect migration
Three details change how you would move a workload onto Neon. First, there is no free tier, so prototypes built on free keys need billing switched on. Second, explicit caching is not supported, only implicit caching at return [ .15 per million tokens; if your current pipeline creates named caches, that code path will not carry over. Third, grounding with Google Search follows the shared rule for Gemini 3 and newer models: 5,000 free search requests per month, then per 1,000 requests.
What to do this week
- If you run Carbon or Argon at the introductory rate, price your December traffic at and , then price the same traffic on Neon at and . The gap is the budget case for a test.
- Run your own eval set on Neon at the default high thinking level and at low. Compare quality and output token counts against the model you use now, since Google publishes no head-to-head figures.
- Move offline jobs (evals, backfills, document processing) to Batch or Flex, where Neon costs half the standard rate.
- Check your code for explicit cache creation and free-tier keys before switching the model id.
For other options across vendors, see the tool directory and the full article archive.
Questions people ask
How much does Gemini 4 Neon cost?
On the standard paid tier, .00 per million input tokens and .00 per million output tokens, with thinking tokens billed as output. Batch and Flex cost .50 and .50, and Priority costs .40 and .00. Google lists no introductory end date for Neon.
Is Gemini 4 Neon free to use?
No. Google's pricing page states that Neon is not available on the free tier. Gemini 4 Hydrogen is the Gemini 4 model listed on both the free and paid tiers.
Is Gemini 4 Neon cheaper than Gemini 4 Carbon?
Not until November 30, 2026, while Carbon is at .00 and .00. From December 1, Carbon moves to .00 and .00, and Neon at .00 and .00 becomes the cheaper of the two on list price.
What is the context window of Gemini 4 Neon?
1M tokens. The changelog describes a unified 1M token context window with long decode up to 1M tokens, and four thinking levels from low to xhigh.
Sources