Google’s Gemini 4 Neon Targets Speed, Cost, and AI Leadership

Google’s Gemini 4 lineup is growing at a pace that is hard to miss. The company has announced Gemini 4 Neon, a model designed to sit between Gemini 4 Hydrogen and the top-end Gemini 4 Carbon, with a focus on combining strong reasoning with faster task completion.
Neon launched on November 18, 2026, across Google’s paid AI services. It is available through the Gemini API in Google AI Studio and Android Studio, while Google AI Pro and Ultra subscribers can use it in the Gemini app and Google Antigravity. Google says Neon brings “frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows.”
A middle model with top-tier ambitions
Google describes Neon as the balanced option in its current lineup. “While Carbon tackles your hardest, long-horizon projects and Hydrogen handles fast everyday execution, Neon balances depth and efficiency,” the company said. That position gives Neon a clear job: handle demanding work without the time and cost tied to Google’s most powerful model.
The early numbers put Neon close to the top. On Google’s own testing, Neon beat Anthropic’s Claude Opus 5.5 and OpenAI’s GPT-6.2 Sol on 16 of 19 benchmarks. Its strongest leads came in long-document work, office automation, and cyber tasks.
Artificial Analysis placed Neon third on its intelligence index with a score of 61. That puts it ahead of Google’s top-end Carbon and behind only Anthropic’s Claude Fable 5.2 and OpenAI’s GPT-6.2 Astra on the listed leaderboard. At its highest setting, Neon also edges past Carbon on the general intelligence index.
The comparison needs some care. The peer results come from the Artificial Analysis leaderboard and the Vals Index, whose public snapshot predates Neon’s release. Reasoning settings also apply to the Artificial Analysis column, and providers define those settings in different ways. Two models running at maximum effort can spend different amounts of time and money on the same question.
Speed and pricing are Neon’s main pitch
Google says Neon matched or exceeded Gemini 4 Carbon on overall quality at three times the speed and 60% of the cost. Thai Tran described the result this way: “In our early evaluations, it matched or exceeded Gemini 4 Carbon on overall quality at 3x the speed and 60% of the cost.”
Neon costs $3 per million input tokens and $15 per million output tokens. Those rates sit 25% below Anthropic’s Claude Opus 5.5, which charges $4 and $20 at its standard rates. Neon’s pricing is also 25% below Carbon’s scheduled rates beginning December 1, although Carbon’s introductory rate makes it cheaper per token through November 30, 2026.
OpenAI’s GPT-6.2 Sol adds another wrinkle. Under its long-context pricing, a Sol request exceeding 272,000 input tokens costs $4 per million input tokens and $15 per million output tokens for the whole request. At that length, Sol loses its headline output-price advantage, while Neon keeps its $3 and $15 rates.
Google’s Vals results show a tighter race among its own models. Hydrogen, Neon, and Carbon all finish within about 2.5 percentage points of one another on the professional tasks measured by Vals. Google supplies Neon’s Vals result, while the public figures include scores of 70.9%, 70.14%, 66.06%, 70.85%, 68.49%, and 70.45% across the listed comparisons.
Where Neon fits in everyday work
Google’s examples focus on tasks that require several types of information at once. A customer could ask Neon to examine utility statements and explain why a charge rose, or cross-check receipts and other records when seeking a refund. Neon can also examine PDFs, spreadsheet entries, and photographs together.
Google demonstrated that broader ability in Sheets by giving a group of Neon agents a single image and asking them to reconstruct it across 100,000 spreadsheet cells. The agents used more than 500,000 formulas to identify edges, map depth, and change the lighting without embedding the original picture or writing a script. The demonstration shows Neon producing a working artifact, not just describing what a user should do.
Thoughtful Lab founder Karina Nguyen said Neon helped her team understand complex software architectures quickly and spotted subtle vulnerabilities that other models missed during the firm’s security evaluations. Cartwheel co-founder and chief scientist Andrew Carr also described using Neon in machine-learning research.
Neon is Google’s fourth Gemini 4 model in under two months. The speed of those releases has drawn attention inside and outside the company. The night before launch, Minkai Xu posted, “many people are surprised by how fast RSI moves,” hinting that self-improving AI may be contributing to Google’s pace.
Google adds more AI features across its products
The Neon launch arrives alongside a series of Google updates. The AI Ultra plan now includes YouTube Premium Family, replacing the earlier individual YouTube Premium benefit. Ad-free YouTube and YouTube Music can be shared with up to five family members, while the cheaper AI Pro plan includes YouTube Premium Lite without YouTube Music benefits.
Google is also expanding access to its Genie 4 world model. After reaching trusted testers late last month, Genie 4 is now available to eligible AI Ultra subscribers in Project Genie. Google has also added temporary threads to AI Mode searches, so people can ask questions without saving them to search history. The company says those sessions will not use past preferences, affect future personalization, or appear in history.
Those temporary chats and chats with history turned off remain available for the last 24 hours of context and system-failure recovery. Google previously held them for 72 hours, but will now keep them for around two months.
Other changes include Google Wallet’s return on iPhone for loyalty cards, gym memberships, payment cards, tickets, and boarding passes from apps such as Gmail. Tap-to-pay is not supported on iOS. Gemini Gems are now read-only and will disappear from personal accounts on December 17th, with Skills becoming Google’s replacement.
Wear OS watches can now use a connected phone’s GPS during outdoor workouts, allowing the watch to switch off its GNSS receiver and save battery. Gemini Notebook can generate interactive sites from notebook sources for users 18 and older, and Google is rolling out 66 regional voices across Android and iOS.
Based on
- Google’s shipping spree continues with Neon — therundownai.beehiiv.com
- Google released another Gemini 4 model. | The Verge — theverge.com
- Google launches Gemini 4 Neon with Opus 5.5-level coding performance at 25% lower API token prices | VentureBeat — venturebeat.com
- Google Archive for November 2026 – Page 1 | The Verge — theverge.com
- Google rolls out Gemini 4 Hydrogen, finally replacing 3.8 Flash | Ars OpenForum — arstechnica.com




