Google released Gemini 4 Neon on Wednesday at $3 per million input tokens and $15 per million output tokens.
The model is rolling out in the Gemini API within Google AI Studio, Android Studio, and the paid tier of the Gemini API through Gemini for business. Paid Google AI Pro and Ultra subscribers also gain access via the Gemini app and Google Antigravity. Input tokens cost $3 per million, while cached input tokens carry a 95 percent discount at $0.15 per million. Output tokens are priced at $15 per million.
Excited to launch Gemini 4 Neon: frontier-level intelligence, super fast task completion, and available to Pro & Ultra users now!! We think you’ll enjoy this model. Alchemy at its finest… 🧪 pic.twitter.com/n9VhNnYKGD
Carbon and Argon carry a promotional rate of $2 per million input tokens and $10 per million output tokens through November 30, before rising to $4 and $20 on December 1. Beginning December 1, Neon’s rates will sit 25 percent below Carbon’s scheduled rates.
In the Gemini app model picker, Neon occupies the deeper analysis option, replacing Argon. Subscribers can select among four thinking levels: Low, Medium, High, and Extra High. The system maintains a context budget of 1,048,576 tokens shared across input, thinking, and output, with up to 65,536 combined thinking and output tokens per synchronous segment.
GOOGLE 🔥: Gemini 4 Neon reference has been spotted on Gemini web! Would be cool if it would be a new Flash Lite model (pure speculation). Looks like Google isn’t done with new releases. * Not available yet 👀 https://t.co/24FVLd08jS pic.twitter.com/dQMqv9PVMZ
Neon is adapted from Gemini 4 Carbon and features a March 2026 knowledge cutoff. The architecture supports a combined input-output context window of up to one million tokens and handles text, images, audio, and video. In multi-step agent workflows, Neon can operate independently or serve as an orchestrator directing parallel Hydrogen subagents inside Google Antigravity.
The development team built an automated reinforcement learning closed loop focused on difficult cross-repository engineering tasks where agents traditionally struggle, placing them in strict sandboxes. For security, the model incorporates real-time protection features and its highest thinking level to resist indirect prompt injection attacks.

Google reported that Neon scored 61 on the Artificial Analysis Intelligence Index at its highest xhigh thinking level, placing third overall. The model secured second place on the Vals Index with a score of 70.9 percent and achieved first place on the Artificial Analysis Coding Agent Index with a score of 58.6.
Google was founded in 1998 by Larry Page and Sergey Brin as a search engine company and reorganized in 2015 as a subsidiary of Alphabet Inc., with Sundar Pichai serving as CEO of both entities.
Gemini 4 Neon: frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows. Begins rolling out today in the Gemini API in @GoogleAIStudio, Gemini for business, for paid Google AI Pro and Ultra… pic.twitter.com/FMiFCFvejw
The company operates the world’s most-visited websites, Google Search and YouTube, and leads global market share in email, mapping, mobile operating systems, web browsers, and AI virtual assistants. Its AI research division, Google DeepMind, developed the transformer architecture that underpins modern large language models.
| Benchmark | Gemini 4 Neon | Claude Opus 5.5 | GPT-6.2 Sol |
|---|---|---|---|
| RiemannBench (science and math) | 84.0% | 80.0% | 80.0% |
| Terminal-Bench-Science | 70.5% | 63.3% | 68.0% |
| HLE-Diamond (reasoning) | 71.3% | 67.9% | 64.0% |
| AA Index v5.0.1 (capability index) | 61 | 59 | 57 |
Neon also scored 65.4 percent in CursorBench 4.1 and reached 39.0 percent on the SWE-sweep test inside the Antigravity environment, alongside 73.5 percent on Chartography and 75.8 percent on Agents’ Last Exam v1.1.
GOOGLE 🔥: Gemini 4 Neon is being prepared for release on Google Cloud. It now appears in the SKU catalog, following similar preparations for earlier releases. Earlier, a reference to Gemini 4 Neon appeared on the Gemini website in the upsell section, suggesting it will require… https://t.co/2SOE6VsYwy pic.twitter.com/mWPWUmXAoY
More on Gemini
Gemini 4 Neon has been added to Antigravity CLI as an Pro orchestrator model. https://t.co/gJFUOuNEqt pic.twitter.com/GOPH11Wqnx
GOOGLE 🔥: Another week, another Gemini model is about to get released! So far we have rumoured Gemini Neon, Helix and Krypton as options as well as a potential Nano Banana upgrade. The next Nano Banana, based on Gemini 4 is the one that was in a preparation for the release… https://t.co/AmPdB5ZyFE pic.twitter.com/kC5iiBDPby
Gemini 4 Neon, the model we caught in the Gemini app on Monday, now has prices in Google Cloud's billing catalog 🔥 • $3 in / $15 out per 1M tokens, live as of 3 AM ET • That puts it between Hydrogen ($1.50 / $7.50) and Carbon ($4 / $20) • Batch and Flex at half price: $1.50… https://t.co/Ugwal0U9LC
GEMINI 4 NEON 👀 A model called gemini-4-neon has surfaced in the Gemini web app's code from Google. For now it's only a code reference, with no public access. pic.twitter.com/737tbgXd1x
Gemini
Starting today, Gemini 4 Neon is available to Google AI Pro and Ultra subscribers on mobile, desktop, and web. This new model excels at work that requires frontier-level reasoning and fast responses. 🧵