Google DeepMind

Gemini 4 Neon

Gemini 4 Neon launched on 2026-11-18 under API id gemini-4-neon, with text, image, video, audio, and PDF input, text output, a 1,048,576-token unified context, and a flat paid Standard rate of $3/$15 per million tokens across that window. Gemini 4 Neon is not on the Free tier. No Gemini 4 Neon score row is stored. Carbon, Argon, Hydrogen, and Gemini 3.8 Flash are not superseded.

Gemini 4 Neon benchmarks and pricing, every number sourced: no Gemini 4 Neon benchmark scores tracked yet, priced at $3.00 per million input tokens and $15.00 per million output.

Released
2026-11-18
License
proprietary
Context window
1M tokensvendor-stated (aistudio.google.com/docs/models/gemini-4-neon: unified context window 1,048,576; checked 2026-11-19)
Knowledge cutoff
2026-03
Parameters
Not disclosed
Architecture
Not disclosed

Gemini 4 Neon API pricing

$3.00 in / $15.00 out per 1M tokens — official pricing source

What Gemini 4 Neon costs per job

Gemini 4 Neon cost for three reference workloads, computed from its list rates
WorkloadTokens in / outCost
One long chat turn100K / 10K$0.450
A codebase review1,000K / 100K$4.50
A day of agent work10,000K / 1,000K$45.00

Computed from Gemini 4 Neon’s list rates above — cache discounts and batch tiers are not applied.

Gemini 4 Neon is one of 8 Google DeepMind models tracked on this site, at these official list prices.

Google DeepMind model pricing, official list rates
ModelIn / 1MOut / 1M
Gemini 4 Neon$3.00$15.00
Gemini 4 Hydrogen$1.50$7.50
Gemini Decision 1$0.10$0.00
Gemini 4 Carbon$4.00$20.00
Gemini 4 Argon$4.00$20.00
Gemini 3.8 Flash$0.75$3.75
Gemini 3.7 Flash(previous version)$0.75$3.75
Gemini 3.1 Pro Preview$2.00$12.00

Notes on the record

Koray Kavukcuoglu's November 18, 2026 blog introduces Gemini 4 Neon, API id gemini-4-neon. Timestamps only, 2026-11-18: @koraykv 21:33:06Z, blog 21:33:07Z, @sundarpichai 21:33:40Z, @GoogleDeepMind 21:34:04Z, @GoogleAI 21:35:30Z, @demishassabis 21:35:24Z, blog 21:35:25Z, @GoogleAIStudio 21:36:09Z, @Google thread 21:51:33Z. Gemini 4 Neon rolls out in the Gemini API and AI Studio, Gemini for business, Antigravity for paid Pro and Ultra, and the Gemini app for Pro and Ultra. Fairwind "Neon for cyber" is an access program, not a catalog model.

The AI Studio page checked 2026-11-19 lists text, image, video, audio, and PDF in, text out, unified context 1,048,576, default output 65,536, and long decode up to 1,048,576. The blog's "1 million-token" wording is not that stored integer. Gemini 4 Neon supports Batch, Flex, Priority, and Thinking, and not image generation, audio generation, or Live API. Cloud knowledge cutoff is March 2026, stored as 2026-03; AI Studio lists none. Parameters and architecture are unpublished, and Gemini 4 Neon's license is proprietary.

Paid Standard is flat $3.00 input and $15.00 output per million tokens, including thinking tokens, implicit cache $0.15, across the whole window. Gemini 4 Neon is not on the Free tier. Batch is 50% of Standard, and explicit caching is unsupported for Gemini 4 models including gemini-4-neon. Flex and Priority are supported, with no Neon dollar table, so those prices are not stored. Search and Maps grounding follow the shared Gemini 3-and-newer schedule, with the request counts in the FAQ.

The Cloud page lists Gemini 4 Neon as GA on November 18, 2026 for verified enterprise customers in select regions, which does not cancel the public rollout above. Implicit caching only, with a 4,096-token minimum on Gemini 4 Neon. Cache writes at 1.25x standard input from January 12, 2027, with TTL for Hydrogen and later, are future cache-control, not today's list price. No Gemini 4 Neon score row is stored. Carbon, Argon, Hydrogen, and Gemini 3.8 Flash are not marked superseded.

Compare with

FAQ

What does Gemini 4 Neon cost?

Paid Standard is $3.00 input and $15.00 output per million tokens, including thinking tokens, across the full 1,048,576-token window. Implicit cache for Gemini 4 Neon is $0.15. Gemini 4 Neon is not on the Free tier, so no free input or output rate is stored. Batch is 50% of that Standard rate. Flex and Priority are supported, and the pricing page publishes no Neon dollar figures for those modes, so none are stored. Checked 2026-11-19.

What is Gemini 4 Neon's context window?

The AI Studio model page checked 2026-11-19 states one unified context window of 1,048,576 tokens for Gemini 4 Neon. Gemini 4 Neon's default synchronous output is 65,536 tokens. Long decode can raise Gemini 4 Neon output to 1,048,576. Explicit caching is not supported for Gemini 4 Neon. This page stores 1,048,576. The blog's "1 million-token" wording is not that stored integer.

Is Gemini 4 Neon on the Free tier?

No. The AI Studio pricing page checked 2026-11-19 says Gemini 4 Neon is not available on the Free tier. The stored price is the paid Standard rate of $3.00 input and $15.00 output per million tokens. Paid Gemini 4 Neon content is not used to improve products. A Free-tier $0 rate is not stored for Gemini 4 Neon.

What can Gemini 4 Neon take in, and what does it return?

Gemini 4 Neon accepts text, image, video, audio, and PDF, and returns text. Image generation, audio generation, and the Live API are not supported on Gemini 4 Neon. Batch, Flex, Priority, and Thinking are supported on Gemini 4 Neon. The Cloud model page names thinking levels LOW, MEDIUM, HIGH, which is the default, and XHIGH. MINIMAL or a thinking_budget on Gemini 4 Neon returns HTTP 400. Those Gemini 4 Neon thinking levels are a vendor description, not a stored score.

What are Gemini 4 Neon grounding rates, and what about Batch, Flex, and Priority?

Paid Google Search grounding for Gemini 4 Neon is 5,000 requests a month free, shared across Gemini 3 and newer, then $14 per 1,000, with a daily cap of 1,500. Maps grounding for Gemini 4 Neon is 5,000 free requests, then $14 per 1,000. The optimization page says Batch is 50% of Standard for Gemini 4 models, including gemini-4-neon. Flex and Priority are supported. No Neon-specific dollar table for Batch, Flex, or Priority is on the pricing page, so those mode prices are not stored. Checked 2026-11-19.

Does Gemini 4 Neon replace Gemini 4 Carbon, Argon, Hydrogen, or Gemini 3.8 Flash?

No. Gemini 4 Neon is a new current Gemini 4 model. The launch materials place Gemini 4 Neon between Carbon and Hydrogen. This page does not mark Carbon, Argon, Hydrogen, or Gemini 3.8 Flash superseded, and it adds no deprecation event for them.

Has Gemini 4 Neon been independently benchmarked?

No. No score row is stored for Gemini 4 Neon. The launch blog states vendor eval claims, including an Artificial Analysis Intelligence Index of 61 at xhigh and a Vals result of 70.9 percent, which the blog ranks second. This scan did not fetch a tracked-board page, so those Gemini 4 Neon claims are not rows.

Where does Gemini 4 Neon run, and what is the knowledge cutoff?

The Cloud model page lists model ID gemini-4-neon, general availability, release date November 18, 2026, and a knowledge cutoff of March 2026, stored here as 2026-03. The AI Studio property table did not list a Gemini 4 Neon cutoff. The Cloud page says Gemini 4 Neon is available to verified enterprise customers in select regions. That is the Cloud surface only. The blog separately describes a public rollout in the Gemini API, AI Studio, Gemini for business, Antigravity for paid Pro and Ultra, and the Gemini app for Pro and Ultra. Fairwind's Gemini 4 Neon for cyber is an access program, not another catalog model.

Further reading