Google Launches Gemini 4 Carbon at Lower Prices Google makes Gemini 4 Carbon generally available, undercutting Anthropic's Claude Fable 5.2 on price while competing closely on performance benchmarks. Google has officially made its Gemini 4 Carbon model generally available across its paid API, business platforms, and consumer products. This launch, which began on November 10, nearly two weeks after the initial October 28 announcement, marks a significant shift in the competitive landscape of frontier artificial intelligence models. The release includes access through the Gemini app, AI Mode in Search, and the Antigravity platform, expanding availability beyond the limited cybersecurity group previously granted access. Pricing Structure and Comparative CostsGoogle DeepMind is aggressively undercutting competitors by pricing Carbon at a fraction of industry standards during the introductory period. From the launch date through November 30, the API costs $2 per million input tokens and $10 per million output tokens. This promotional rate is roughly one-quarter of the standard pricing for Anthropic's Claude Fable 5.2, creating a substantial cost advantage for developers and enterprises looking to migrate workloads. After November 30, Google's prices will double to $4 for input and $20 for output, which still represents half of Claude's standard rates. Additionally, the company offers a ninety-five percent discount on cached inputs, bringing the cost down to $0.10 per million tokens during the promotion.Performance Benchmarks and Technical MetricsThe performance comparison between Gemini 4 Carbon and Claude Fable 5.2 reveals distinct strengths for each model depending on the specific task at hand. On general intelligence benchmarks, Claude Fable 5.2 maintains a lead with a score of sixty-five on the Artificial Analysis Intelligence Index, compared to Carbon's score of sixty. Fable also outperforms on code-review tasks such as FrontierCode 2.0, achieving a forty-nine point one percent success rate against Carbon's forty-point six percent. However, Carbon excels in agent-focused scenarios, scoring fifty-three point eighty-eight percent on Zapier's AutomationBench, significantly higher than Fable's fifty-one point four five percent. Google claims Carbon leads on agent benchmarks like CursorBench 4.0 while costing less than half of the next best alternative.Consumer Subscription Plans and AccessFor individual consumers, the new model is being rolled out through specific subscription tiers designed for heavy AI usage. Google AI Ultra subscribers can now access the model at $99.99 per month for the Ultra 5x tier or $199.99 per month for the Ultra 20x tier. This consumer rollout coincides with the broader enterprise availability, allowing a wider audience to interact with the model through Google's search interface and standalone applications. The timing is strategic, occurring just one day after Anthropic launched its own Claude Fable 5.2 and Mythos 5.2 models, intensifying the direct competition for market share in the generative AI sector.Enterprise Adoption and Total Cost AnalysisThe decision to offer such steep discounts suggests Google is prioritizing rapid enterprise adoption and workload migration over immediate revenue maximization. While token prices are low, the actual cost of completing complex tasks varies based on effort levels and retry rates. Artificial Analysis estimates that completing tasks at high reasoning levels costs an average of $7.76 for Carbon, compared to $26.54 for Claude at maximum effort. However, when adjusting for lower effort levels, the cost gap narrows significantly, with Fable costing $3.80 at medium effort. Enterprises must weigh these factors carefully, considering that the least expensive configuration depends heavily on the specific quality targets and failure rates of their internal workflows. The promotional window ends on November 30, after which the pricing structure will adjust but remain competitive against standard market rates.