Key Notes
- Gemini 4 Carbon leads Artificial Analysis’s Intelligence Index v5, while initial access remains limited to selected cyber defenders.
- Introductory API prices of $2 per million input tokens and $10 per million output tokens will later double.
- The launch adds competitive pressure as Alphabet funds its infrastructure buildout and Anthropic prepares a proposed Nasdaq IPO.
Google unveiled Gemini 4 Carbon on October 28, strengthening its position in the frontier AI race just as investors scrutinize the cost of that competition. The new model leads Artificial Analysis’s latest overall intelligence index, but its initial rollout is restricted to selected cybersecurity defenders.
For parent Alphabet Inc. (NASDAQ: GOOGL, GOOG), the announcement connects a technical advance with a financial question: how quickly can better models turn rising infrastructure investment into durable revenue and cash returns?
Google’s announcement sets introductory API prices of $2 per million input tokens and $10 per million output tokens. Broader availability for developers, businesses and consumers is planned, but no general release date has been announced.
A Benchmark Lead With a Limited Initial Rollout
Artificial Analysis places Carbon first on its Intelligence Index v5 with a score of 59. That is an index value, rather than a claim that the model completes 59% of every type of task. The evaluator’s assessment identifies gains in reasoning, scientific work and coding compared with Google’s earlier models.
The lead is not universal. On the separate Coding Agent Index, Carbon scores 59 using Antigravity CLI, while Anthropic’s Claude Opus 5.5 scores 62 in Claude Code. These results compare particular model-and-tool configurations, making the workflow relevant alongside the underlying model.
Google is starting access with a subset of cyber defenders and plans to expand through its Fairwind partner program. The controlled introduction reflects the sensitivity of capabilities that can find and repair vulnerabilities, but can also be misused.
AIstify’s technical breakdown examines Carbon’s benchmarks, pricing and access restrictions in detail. For investors, those restrictions mean the announcement should not be treated as an immediate, unrestricted commercial launch.
Pricing Puts the Economics of AI Work in Focus
Carbon’s introductory rates match Gemini 4 Argon’s advertised launch pricing. After the introductory period, Google says input tokens will cost $4 per million and output tokens $20 per million. It has not specified when Carbon’s introductory period ends.
Cached inputs receive a 95% discount from the applicable input rate. That implies $0.10 per million cached input tokens initially and $0.20 at standard pricing. These are usage prices, not a fixed cost for completing a business assignment.
A stronger model can improve the economics of a workflow if it requires fewer retries, less supervision or less time to produce an accepted result. Longer reasoning runs, additional tool calls and review requirements can also increase the total bill. A lower token rate does not automatically establish a lower cost per completed task.
That is important for competition with Claude and other enterprise systems. Customers will assess reliability, integration and total spending, while providers must balance demand against the computing resources required to serve it. Google has not disclosed a Carbon-specific revenue forecast or profit margin.
Alphabet’s Earnings Expose the Spending Burden
Alphabet’s latest financial results show both the scale of the opportunity and the cost of pursuing it. Third-quarter revenue increased 25% to $127.58 billion, while Google Cloud revenue rose 103% to $30.71 billion.
Cloud’s growth reflects enterprise AI infrastructure, AI solutions and core platform services, with TPU-system sales also contributing. The reported total therefore extends beyond payments for Gemini model usage and should not be read as Carbon revenue.
The cash-flow comparison is more demanding. Alphabet generated $42.29 billion in operating cash flow but spent $51.36 billion on property and equipment, leaving quarterly free cash flow negative by $9.07 billion. Those are companywide figures, not a disclosed budget for training or running Carbon.
As we reported, Alphabet shares fell more than 10% in the initial October 28 after-hours reaction to the earnings release. That response reflected the wider results and investment outlook; it does not establish a separate market verdict on Carbon.
Anthropic’s IPO Faces a Moving Competitive Target
The announcement also arrives as Anthropic prepares its planned Nasdaq listing under the proposed ticker ANT. Our IPO guide examines the offering process, Claude’s revenue growth, financing arrangements and the risks prospective shareholders would inherit.
A competing model’s improvement can affect the assumptions used to value an AI provider. Investors need to consider how long performance advantages last, whether customers remain loyal as alternatives improve and how much pricing power a provider can retain.
Carbon’s overall index lead does not demonstrate that Anthropic will lose customers, and Claude’s coding-agent result shows that important competitive strengths remain. The financial issue is the durability of differentiation across the tasks companies actually pay to complete.
Alphabet and Anthropic also offer different investment exposures. Alphabet shareholders own a diversified group with advertising, cloud and other businesses; an eventual Anthropic shareholder would have a more direct exposure to Claude’s commercial success and the capital required to support it.
Deployment and Margins Will Decide the Investment Case
Google says Carbon can sustain demanding engineering work over long periods and reports strong cybersecurity results. Those are capability claims under specified evaluations and internal use, rather than proof of equivalent reliability in every customer environment.
The next commercial checkpoints are broader access, production deployments, repeat usage and evidence that customers obtain useful results at an acceptable cost. For Alphabet, those developments must eventually support revenue and margins sufficient to justify the infrastructure supporting them.
Carbon strengthens Google’s technical offering. Its investment significance will become clearer as deployment expands and the economics of paid use emerge, against the spending already visible in Alphabet’s accounts and the expectations surrounding Anthropic’s proposed IPO.
