24 Sources
[1]
Google announces Gemini 4 Carbon, its latest frontier AI model that you can't use
Gemini 4 Carbon is essentially a slightly better version of Argon. It's almost like Google's got something to prove after taking a break from releasing new frontier AI models earlier this year. Just two weeks after rolling out Gemini 4 Argon, Google has announced Gemini 4 Carbon. This isn't the
[2]
Google Debuts New AI Models, Genie 4 and Gemini 4 Carbon: What You Need to Know - CNET
The world model has a waitlist and the newest frontier Gemini model is restricted to select cybersecurity experts. Google is back with another new frontier AI model, just two weeks after its last release. The new large language model is dubbed Gemini 4 Carbon, the company announced on Wednesday.
[3]
Google launches Gemini 4 Carbon with a 'significant leap' in cybersecurity
Google has released its latest frontier model, Gemini 4 Carbon, offering "significant gains" in coding, agentic tasks, and cybersecurity. In an announcement on Wednesday, Google highlights Carbon's top spot on the Artificial Analysis Intelligence Index (v5.0) benchmark, with Claude Opus 5.5 and
[4]
Google reveals new Gemini 4 Carbon AI model, narrowly beating Claude Opus 5.5 -- frontier AI model is also the least likely to hallucinate, benchmarks claim
Last week, Google released its latest AI model, Gemini 4 Carbon, topping intelligence rankings charts. The model has landed with a grander splash than most. As Artificial Analysis highlights, Gemini 4 Carbon is now the highest-ranked AI model on its latest Intelligence Index V5 benchmark suite,
[5]
With Gemini 4 Carbon, Google Touts Big Leap in Cybersecurity Capabilities
After launching the Gemini 4 Argon model earlier this month, Google released the Gemini 4 Carbon model on Wednesday, which it says is built for sustained agentic workflows and delivers a "significant leap in cybersecurity capabilities." Testing shows Carbon outperforming Argon, OpenAI's GPT-6
[6]
Google's Gemini 4 Carbon looks like a diamond in the rough
Just two weeks after launching Gemini 4 Argon, Google has announced a better-performing version of the model that it will sell for the same price, at least temporarily. Koray Kavukcuoglu, SVP at Google DeepMind and chief AI architect at Google, said Gemini 4 Carbon represents the company's
[7]
Google's Gemini 4 Carbon Leads Benchmarks for Finding, Exploiting, and Patching Vulnerabilities
Google on Wednesday took the wraps off Gemini 4 Carbon, its latest frontier model with state-of-the-art cybersecurity and long-horizon capabilities. "It shows significant gains across coding and agentic tasks as well as across key capabilities critical for cybersecurity defense, landing a firm
[8]
Gemini 4 Carbon tops AI rankings, and cyber defenders get it first
Gemini 4 Carbon scores 88.2% on a test of finding real vulnerabilities, Google says, and beats Argon at writing secure code. Rivals still lead two of Google's own benchmarks. Google has released Gemini 4 Carbon, its newest frontier AI model, to a small group of cyber defenders. It arrives two
[9]
Gemini 4 Carbon is Google's second new frontier model this month
Carbon will be available at $2 per million input tokens and $10 per million output tokens, with cached input tokens priced at 95% off the input token price. About a month ago, Google teased us with its next frontier model -- Gemini 4 Argon. Two weeks ago, that model saw the light of day as it
[10]
Google's Gemini 4 Carbon AI finds 88.2% of security flaws
The AI model will initially be available to selected cybersecurity defenders before a broader rollout. Google has introduced Gemini 4 Carbon, a new artificial intelligence model designed to identify software vulnerabilities, generate security patches, and handle complex engineering tasks that can
[11]
Gemini 4 Carbon lags just behind Claude in benchmark, but wins in value
Google's new frontier model, Gemini 4 Carbon, brings some welcome improvements and while it doesn't top the charts in a new benchmark, it does provide much more bang for your buck. SemiAnalysis, an AI research and analysis firm, recently published its new frontier-level benchmark, GDPx which
[12]
Google announces Gemini 4 Carbon, its new frontier AI model
Google announced Gemini 4 Carbon, a new frontier AI model that arrives just two weeks after Gemini 4 Argon reached the Gemini app. In a blog post published Oct. 28, Koray Kavukcuoglu, SVP at Google DeepMind, the company's frontier AI lab, called Carbon Google's strongest model yet. (Kavukcuoglu is
[13]
Google unveils Gemini 4 Carbon, retaking the world's no. 1 spot on updated, harder benchmarks -- but access is limited
Google announced its latest and most intelligent frontier AI model today: Gemini 4 Carbon promises improvements in cybersecurity and complex software engineering at the same introductory token prices as its previous flagship, Argon. It also claims the top spot in the world on most common enterprise
[14]
Google's new AI model Gemini 4 Carbon is built for complex coding and cybersecurity tasks
Gemini 4 Carbon currently ranks first on Artificial Analysis' updated Intelligence Index. Google just dropped its latest AI model, Gemini 4 Carbon, designed to tackle complex coding jobs and cybersecurity threats. The company claims it can work through difficult engineering problems for days,
[15]
Google debuts flagship Gemini 4 Carbon model with cybersecurity upgrades
Google LLC today debuted Gemini 4 Carbon, a frontier large language model adept at complex programming tasks. The LLM is rolling out just two weeks after the search introduced its previous flagship LLM, Gemini 4 Argon. According to Google, internal tests have shown Carbon to be more than twice as
[16]
Google unveils Gemini 4 Carbon for coding and cybersecurity
Google introduced Gemini 4 Carbon, an AI model it says improves coding and cybersecurity performance, with initial access limited to selected cybersecurity defense companies for pre-release testing. Google plans to extend access to additional trusted partners through its Fairwind Program before
[17]
Google's Frontier Model is Cost Efficient but Not the Best at Agentic Tasks
Claude Fable 5.1 was the least cost-efficient among the AI models Google launched its latest frontier AI model late last month, named Gemini 4 Carbon, with agentic and enhanced cybersecurity capabilities. The AI model was released just two weeks after the Mountain View-based tech giant released
[18]
Google Gemini 4 Carbon Beats Claude Opus 5.5 in 16 of 18 Tests
Google's latest AI model, Gemini 4 Carbon, builds on the foundation of Gemini 4 Argon, introducing capabilities tailored to address pressing challenges in cybersecurity and task management. In a detailed breakdown by Paul Lipsky, the model's standout features are explored, including its ability to
[19]
Elon Musk Has a One-Word Verdict on Google's Gemini 4 Carbon - Alphabet (NASDAQ:GOOGL), Alphabet (NASDAQ:
Alphabet (NASDAQ:GOOGL)'s Google is rolling out its new Gemini 4 Carbon frontier AI model, and Space Exploration Technologies (NASDAQ:SPCX) CEO Elon Musk said he finds it "impressive." Gemini 4 Carbon touts a top spot on Artificial Analysis' Intelligence Index and strong cybersecurity benchmark
[20]
This is Google's New Frontier AI Model With Cybersecurity Capabilities
Gemini 4 Carbon is currently available for select cyber defenders Gemini 4 Carbon was introduced by Google on Wednesday as its latest frontier AI model. The Gemini 4 Carbon model arrives weeks after the tech giant released the Gemini 4 Argon. Google's latest frontier model is currently available
[21]
Google introduces Gemini 4 Carbon with improved coding and cybersecurity capabilities
Google has announced Gemini 4 Carbon, its latest frontier AI model, building on Gemini 4 Argon, which launched two weeks ago. Gemini 4 Carbon Carbon ranks first on the recently updated Artificial Analysis Intelligence Index v5.0 with a score of 59. Google is beginning with a limited rollout to
[22]
Gemini 4 Carbon: 9 Breakthroughs Transforming AI Coding and Cybersecurity
Next-Generation AI Model: Google unveiled Gemini 4 Carbon on October 28, 2026, positioning it as its strongest model for advanced reasoning, software engineering, and cybersecurity. . Advanced Cybersecurity Detection: Carbon achieved 88.2% on SEC-bench Pro, demonstrating its ability to identify
[23]
Google launches Gemini 4 Carbon AI model with cybersecurity focus
Oct 28 (Reuters) - Alphabet's Google on Wednesday unveiled Gemini 4 Carbon, its latest frontier AI model, stepping up competition with OpenAI and Anthropic as the race to build more capable coding and cybersecurity systems intensifies. The launch comes two weeks after Google introduced Gemini 4
[24]
Google unveils Gemini 4 Carbon AI model, says it delivers major leap in cybersecurity
Google has unveiled Gemini 4 Carbon, its latest AI model with improved coding, reasoning and cybersecurity capabilities. The company says the model delivers a significant improvement in finding software vulnerabilities, testing potential attacks and fixing security flaws. Google claims Carbon has
Share
Copy Link
Google launched Gemini 4 Carbon just two weeks after Gemini 4 Argon, claiming it as the most capable frontier AI model for cybersecurity. The model tops benchmarks against Claude Opus 5.5 and GPT-6 Astra while resisting all prompt injection attacks in testing. Access remains limited to select cyber defenders through Google's Fairwind Program.

Google has launched Gemini 4 Carbon, its latest frontier AI model, just two weeks after releasing Gemini 4 Argon
1
2
. The announcement signals Google's aggressive push to regain momentum in the AI industry after an 11-month gap without releasing new frontier models earlier this year2
. The rapid succession of releases demonstrates the company's renewed commitment to competing with rivals like OpenAI and Anthropic, though the model's restricted availability reflects growing industry caution around powerful AI systems.Gemini 4 Carbon represents an improvement over Gemini 4 Argon across multiple dimensions, though the gains are incremental rather than transformative
1
. Google positions Carbon as a specialized tool for sustained agentic tasks, with internal teams reporting the model running for days to complete complex, multi-step engineering challenges3
. This persistence through obstacles marks a shift toward AI systems capable of handling extended workflows without human intervention.The Google AI model delivers what the company calls a "significant leap" in cybersecurity capabilities, particularly in vulnerability detection, validation, and remediation
1
3
. Carbon generates code twice as likely to be vulnerability-free compared to Argon, while Google's Chrome team reports the model produces patches 10 times better quality than non-Google models1
. These improvements matter as organizations struggle with the PHP vulnerabilities that recently affected a majority of the world's websites2
.On specialized cybersecurity benchmarks, Gemini 4 Carbon holds the number one position on SEC-bench Pro, which measures AI's ability to discover real-world bugs, and ExploitGym, which evaluates how models turn vulnerabilities into exploit generation
3
. The model also scored 89 points higher on WebDev Arena compared to previous versions, now ranking fourth on that leaderboard1
.Gemini 4 Carbon achieved the top score on the Artificial Analysis Intelligence Index V5 benchmark, scoring 59 points—one point ahead of Claude Opus 5.5 and two points ahead of GPT-6 Astra
4
. The model secured first place across four of nine evaluations while maintaining a lower cost per task at $7.76 compared to competitors4
. However, the narrow margins between top models—often just a few points—suggest the frontier AI model competition has reached a plateau where meaningful differentiation proves difficult1
.More significantly, Carbon topped the AA-Omniscience v1.1 test measuring hallucination rate and knowledge recall with a score of 54 on a -100 to +100 scale
4
. Google claims Carbon cuts unsupported claims by two times compared to Argon3
. Yet even the best score of 54 reveals that frontier models in 2026 remain far from consistently reliable when recalling information accurately4
.In security evaluations, Gemini 4 Carbon achieved a 0.0% attack success rate on the Grey Swan prompt injection benchmark, outperforming Gemini 4 Argon at 0.7%, Claude Opus 5.5 at 1.0%, and GPT-6 Astra at 8.5%
1
5
. Prompt injection attacks represent a critical vulnerability where malicious actors manipulate AI systems to bypass safeguards and execute unintended actions5
. Carbon's perfect resistance record positions it as the most secure option for organizations concerned about AI system exploitation.The model demonstrates improved transparency by refusing to complete actions when it lacks capability rather than attempting workarounds
4
. This honest acknowledgment of limitations contrasts with earlier models that would generate plausible-sounding but incorrect responses.Related Stories
Google is implementing a deliberate, phased release strategy for Gemini 4 Carbon, starting with a subset of cyber defenders before expanding to trusted partners through its Fairwind Program
3
5
. The company is actively engaged with the U.S. government's voluntary pre-release model testing framework, similar to Anthropic's approach with previous releases4
. This cautious rollout reflects growing concerns about AI contributing to job losses, misinformation, and deepfakes2
.Google declined to specify when broader access would open to AI Ultra subscribers, developers using the Gemini API, and general consumers
2
5
. Based on Argon's timeline, access could expand within one to two weeks2
. The company wants to iterate on guardrails before making Carbon available to developers, specifically to prevent nefarious actors from exploiting its cybersecurity capabilities for hacking4
.Gemini 4 Carbon will launch at an introductory price of $2 per million input tokens and $10 per million output tokens—matching Argon's initial rates
1
3
. These prices will double after the promotional period expires, though Google hasn't specified the duration1
. Argon's discounted rates end in November1
.Alongside Carbon, Google announced Genie 4, a world model capable of simulating real and imagined physical environments
2
. Unlike chatbots that predict text, world models operate based on digital physics, making them valuable for robotics and video game design2
. Genie 4 introduces audio capabilities, allowing developers to add background noise and dialogue to AI-generated worlds2
. A waitlist is now open for developers interested in accessing Genie 42
. Watch whether Google can maintain this release velocity and whether Carbon's cybersecurity advantages translate into real-world adoption among enterprises prioritizing security over raw performance.Summarized by
Navi
[4]
01 Oct 2026•Technology

10 Nov 2026•Technology

08 Oct 2026•Technology
