BTC $62,614.24 -3.21%
ETH $1,768.33 -2.65%
BNB $579.87 -2.82%
XRP $0.9619 -2.05%
SOL $73.16 -2.65%
TRX $0.3096 +0.65%
DOGE $0.0585 -2.29%
ADA $0.1506 -2.65%
BCH $207.54 -1.92%
LINK $9.02 -2.65%
HYPE $66.48 -4.58%
AAVE $108.89 -3.19%
SUI $0.7160 -2.58%
XLM $0.1363 -3.56%
ZEC $906.13 -6.08%
AAPL $353.92 +0.89%
AMZN $270.43 +4.09%
GOOGL $285.92 +0.94%
MSFT $512.53 +0.57%
META $620.19 -0.42%
NVDA $152.09 -5.82%
TSLA $385.76 +0.59%
SNDK $783.48 -5.79%
INTC $55.72 -0.78%
SPCX $144.87 +1.24%
MU $714.87 -3.66%
AMD $535.35 +2.78%
BTC $62,614.24 -3.21%
ETH $1,768.33 -2.65%
BNB $579.87 -2.82%
XRP $0.9619 -2.05%
SOL $73.16 -2.65%
TRX $0.3096 +0.65%
DOGE $0.0585 -2.29%
ADA $0.1506 -2.65%
BCH $207.54 -1.92%
LINK $9.02 -2.65%
HYPE $66.48 -4.58%
AAVE $108.89 -3.19%
SUI $0.7160 -2.58%
XLM $0.1363 -3.56%
ZEC $906.13 -6.08%
AAPL $353.92 +0.89%
AMZN $270.43 +4.09%
GOOGL $285.92 +0.94%
MSFT $512.53 +0.57%
META $620.19 -0.42%
NVDA $152.09 -5.82%
TSLA $385.76 +0.59%
SNDK $783.48 -5.79%
INTC $55.72 -0.78%
SPCX $144.87 +1.24%
MU $714.87 -3.66%
AMD $535.35 +2.78%

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

Core Viewpoint
Summary: Google released the "medium cup" Neon of the Gemini 4 family: speed is 3 times that of Carbon, cost is 60%, $3 per million inputs / $15 per million outputs, 1 million token context; Artificial Analysis ranked 3rd, with multiple firsts in legal, tax, code, and security benchmarks. The world model Genie 4 was made available to Ultra users starting on the same day.
Recommended Reading
2026-11-19 08:07:48
Google released the "medium cup" Neon of the Gemini 4 family: speed is 3 times that of Carbon, cost is 60%, $3 per million inputs / $15 per million outputs, 1 million token context; Artificial Analysis ranked 3rd, with multiple firsts in legal, tax, code, and security benchmarks. The world model Genie 4 was made available to Ultra users starting on the same day.

Editor | Cold Cat

After waking up, Google has once again released the Gemini 4 model.

Google has been focusing on foundational models recently, continuously releasing models that range from maximum intelligence to maximum efficiency, truly striving to reclaim its throne.

After Carbon, Argon, and Hydrogen, the "medium cup" model Gemini 4 Neon has arrived, filling the gap between maximum intelligence and maximum efficiency!

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

Demis Hassabis stated that the Gemini 4 Neon medium cup model is "Alchemy at its finest."

In the Gemini 4 family, Carbon is the flagship, responsible for the most challenging long-range projects; Hydrogen is aimed at everyday high-frequency scenarios, focusing on response speed. Neon is positioned in between, "providing cutting-edge reasoning capabilities in complex programming, analytical reasoning, and production-grade enterprise workflows, while being faster and more economical."

Neon is three times faster than Carbon, with costs only 60% of Carbon's. The pricing is $3 per million input tokens, $15 per million output tokens, with cached input tokens as low as $0.15, a context window of 1 million tokens, and it also supports four adjustable levels of thinking depth, allowing developers to flexibly switch speed and reasoning depth based on task complexity.

Gemini 4 Neon runs for 12 hours per swarm, reconstructing Alan Turing's research from 1936 from archival photos ------ including a functioning universal Turing machine, the blueprint of modern computing.

Blog link:

Comprehensive Scoring

Neon performs quite comprehensively in third-party evaluations. In terms of overall intelligence, it ranks 3rd with a score of 61 on the Artificial Analysis Intelligence Index, only behind larger, slower models; it ranks 2nd on the Vals Index (70.9%), sitting on the Pareto frontier.

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

In the brand new Agent Arena Index, Neon secured 2nd place (1336 points) with over 4200 real-world agent conversations, and its steerability also ranks 2nd, meaning the model can adjust direction more cleanly when users change their minds midway.

Evaluations for enterprise and knowledge work categories have also seen significant rankings. Legal Research Bench and Tax Agent Bench both ranked 1st, while various categories covered by Surge AI, including finance, accounting, and spreadsheet modeling, mostly ranked 1st as well. Zapier AutomationBench ranked 1st (56.3%), and Andon Labs' Vending-Bench 2 also ranked 1st ($16,843), the latter being a test simulating long-term business operations, which requires high long-range planning capabilities. The intensive chart reading benchmark Chartography scored 73.5%, and Agents' Last Exam v1.1 scored 75.8%, on par with leading models.

In the fields of scientific and mathematical reasoning, Neon achieved first place in Terminal-Bench Science (70.5%), RiemannBench (84.0%), and HLE-Diamond (71.3%), all being the best in their class.

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

Breakdown of Neon Capabilities

According to the official blog, Neon’s new capabilities can be summarized into several major directions.

Writing code, more accurate and stable. Neon is aimed at real production environments, capable of handling multi-file debugging, refactoring, and precise editing. Google specifically mentioned its "surgical edits" capability, which allows for pinpoint modifications in multi-file projects. The model shows significant improvement in following multi-step instructions, and recorded the lowest fabrication rate to date in internal Google Antigravity tests.

1 million token window achieves multimodality. Spreadsheets, PDFs, charts, audio, video, Neon can simultaneously digest these inputs within a single context window, outputting structured deliverables. "Neon matches or even exceeds Carbon in overall quality, is three times faster, and reduces costs by 40%. It finds a convincing balance between accuracy, detail, and speed."

Cybersecurity direction. This is a unique move in Neon’s release. Google launched Neon for Cyber, covering the complete chain of secure code writing, vulnerability discovery, vulnerability validation, and vulnerability remediation.

From the data, AutoBaxBench's secure code generation reached SOTA (76%) at the difficult level, significantly surpassing Carbon; WebExploitBench v1.2 penetration testing also achieved SOTA (51.26%), with costs only half that of competing large models; SEC-bench Pro vulnerability discovery SOTA (90.5%), with single challenge costs about half of similar models; CWE-bench v2 vulnerability remediation at 70.7%, improving 3.4% over Carbon. Neon for Cyber will be opened to trusted security defenders through the Fairwind Program in the coming days.

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

Security alignment, almost "zero cheating." Google has made significant investments in security. Terminal-Bench 4.0 has a cheating rate of 0.0%, on par with Carbon; in coding cheating assessments, only 2.45% of tasks exhibited shortcut behaviors, the lowest across the entire Gemini 4 series; the misconduct rate in the agent security suite is 0.09%, with an honest assessment pass rate of 99.2%. In the indirect prompt injection benchmark of Gray Swan, Neon achieved a 0.0% attack success rate at the highest thinking level. Compared to Carbon, Neon reduced "service insufficiency" by 43%-56% and "overreach" by 15%-35%. At the same time, it almost never refuses legitimate tasks, with only a 1.98% refusal rate on Terminal-Bench 4.0 and a refusal rate of 0% on Terminal-Bench-Science.

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

Consumer-side upgrades. In the Gemini app, Neon brings higher quality writing output, can read uploaded content such as spreadsheets, PDFs, and photos more accurately, and can access Gmail and Docs for relevant information. The Tasks feature is now powered by Neon, making it faster and more economical when handling complex multi-step tasks.

Gemini 4 Neon has begun rolling out. Developers can build applications through the paid tier of the Gemini API in Google AI Studio and Android Studio, while Google AI Pro and Ultra subscribers can try it out in Google Antigravity. Enterprise users can access the Gemini for Business API or model dropdown menu. For consumers, Pro and Ultra subscribers can directly switch to Neon in the model selector of the Gemini app. The cybersecurity-focused Neon for Cyber will be pushed to trusted defenders through the Fairwind Program in the coming days.

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

World Model Genie 4 Finally Launched

On October 28, Google DeepMind officially announced the latest generation world model Genie 4, calling it "the most realistic and interactive world model to date."

After more than half a month of waiting, Google finally announced that Genie 4 is officially available.

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

Blog link:

Genie 4 is Google's most realistic and interactive world model to date. Its core new capability is "mid-session prompting," which allows you to change the environment using natural language at any time while exploring the world. You can switch the weather, introduce a character, or change the plot direction, with all changes occurring in real-time on the current screen.

Another newly introduced feature is native audio. The world generated by Genie 4 comes with sound, and the sound automatically adapts to changes in the scene, including environmental sound effects and character dialogues, filling in the "auditory" puzzle.

Genie 4 focuses on the highest fidelity, while Genie 3 continues to be available in the Project Genie model selector, aimed at longer explorations and panoramic scenes.

With the launch of Genie 4, Google also introduced Project Genie Agent. Previously, creating a world required you to carefully organize prompts, but now it has turned into a conversation. You describe the world you want in natural language, and the agent will orchestrate Google's multimodal model combinations to generate initial visual images to establish the visual style of the world, along with custom background music.

Interestingly, the Gemini model can observe the screen content while you explore the world, actively popping up suggestions for "where to go next," or automatically triggering events within the world based on your journey.

You can also remix existing worlds in the gallery, adjusting the underlying structure, soundtrack, or rules to turn it into your own version, and then publish it to the community gallery for others to explore and create anew.

Currently, the new experience of Project Genie is open to eligible Google AI Ultra subscribers, and Google AI Pro members can join the waiting list on the same page.

Google Gemini 4 Neon released: speed 3 times, cost 60%, world model Genie 4 synchronized open

ChainCatcher reminds readers to view blockchain rationally, enhance risk awareness, and be cautious of various virtual token issuances and speculations. All content on this site is solely market information or related party opinions, and does not constitute any form of investment advice. If you find sensitive information in the content, please click , and we will handle it promptly.

app_icon
ChainCatcher Building the Web3 world with innovations.