Products & Models · Nov 23, 2026
Google doubles API prices for its top Gemini models from December 1, including batch processing
Discounts on Gemini 4 Carbon and Argon end November 30. Standard, Batch, Flex and Priority rates all double, so budgets should be revised on the assumption that monthly costs will double
Rie Suzuki · Technology Editor

Key points
- API prices for Gemini 4 Carbon and Argon double starting with usage on December 1. The current discounted prices apply through November 30
- Standard rates per 1 million tokens rise from $2 (about ¥318) for input and $10 (about ¥1,590) for output to $4 (about ¥636) and $20 (about ¥3,180). Batch, Flex and Priority rates also double
- At the same usage, monthly bills will double. Companies should identify this month any workloads that could move to the mid-tier Neon ($3/$15) or the lower-tier Hydrogen ($1.50/$7.50)
Google will double the price of using its top AI models, Gemini 4 Carbon and Gemini 4 Argon, through its API, starting with usage on December 1. According to Google's Gemini API pricing page, standard rates are not the only ones going up. The discounted bulk-processing tier (Batch), Flex, which runs jobs when capacity is free, and Priority, which puts urgent requests at the front of the queue, will all double as well. Whatever tier a customer uses, the same volume of usage will cost twice as much each month.
For companies in Japan: Companies that call Gemini 4 Carbon or Argon through the API, whether in their own systems or in tools built by outside contractors, should have IT staff and accounting check model names and monthly usage on their billing statements by November 30. They should then either rebuild budgets from December on at double the current level or move to cheaper models. Companies that do not use the Gemini API, and companies that only have monthly per-seat subscriptions to the Gemini app or Google Workspace, are not directly affected by this change to the API price list.
Gemini 4 top-model price increase takes effect December 1, not November 30
The price increase for Gemini 4 Carbon and Argon takes effect on December 1. Nohumans previously reported that prices would change on November 30. According to Google's pricing page, the current discounted prices apply through November 30, and the new prices apply to usage from December 1. November 30 is the last day the discounted prices are available.
The date of the Gemini 4 Argon price increase had not previously been confirmed. For Gemini 4 Carbon, Google itself had stated since the model became generally available on November 10 that the discounted pricing would run through November 30. Google's pricing page has now confirmed that Argon will see the same increase from December 1.
New prices for Gemini 4 Carbon and Argon: input and output both double
Standard rates for Gemini 4 Carbon and Argon per 1 million tokens rise from $2 (about ¥318) for input and $10 (about ¥1,590) for output to $4 (about ¥636) for input and $20 (about ¥3,180) for output. Yen amounts are approximate, calculated at ¥159 to the dollar (the European Central Bank reference rate on November 20). Tokens are the units AI models use to count text; in Japanese, one character is roughly one token.
Batch, Flex and Priority rates on the Gemini API also double for Gemini 4 Carbon and Argon. Batch is a discounted rate for submitting large volumes of non-urgent work together, and many companies use it for overnight summarizing or classifying of documents. Companies that kept costs down with Batch will not escape the increase. Priority is a rate that ensures requests are processed first even at busy times, and is used for things such as customer-facing chatbots.
The increase strips Google's top models of their price advantage. The new $4/$20 is still half the $8/$40 charged for OpenAI's GPT-6.2 Astra and Anthropic's Claude Fable 5.2. But it is now more expensive than Google's own mid-tier model, Gemini 4 Neon ($3/$15).
Budgets for companies using Gemini through the API: plan for monthly costs to double
Companies using Gemini 4 Carbon or Argon through the API should, to be safe, estimate their monthly costs from December at twice current levels. According to Google's pricing page, every tier — standard, Batch, Flex and Priority — doubles, so there is no way out such as switching to Batch to absorb the increase. For example, a company paying ¥100,000 a month for Carbon this month would pay about ¥200,000 in December for the same volume.
Companies looking to contain costs after the increase can reassign models task by task. In November, Google released the mid-tier Gemini 4 Neon ($3/$15 per 1 million tokens) and the fast, low-cost Gemini 4 Hydrogen ($1.50/$7.50). By Google's own figures, Hydrogen outperforms Argon on an overall index, and placing it in a role working under Carbon cuts usage by about 24%. For tasks such as sorting customer inquiries or routine summarization, it is worth testing whether a lower-tier model is sufficient. Designing the harness — the system that bundles the instructions, tools and procedures given to an AI — so that only difficult decisions are routed to Carbon can reduce the impact of the increase.
The increase could also affect the prices charged by outside development contractors and by business services with built-in AI. Even companies that do not contract for the API directly may see providers pass the increase on to customers if Carbon or Argon runs behind the services they use. Companies with contract renewals in the middle of the fiscal year should ask their providers which models they use.
What could not be confirmed about the Gemini 4 price increase
For Gemini 4 Carbon and Argon, Batch and Flex rates per 1 million tokens rise from $1 for input and $5 for output to $2 and $10, and Priority rates rise from $3.60 for input and $18 for output to $7.20 and $36. Cached input (a discount for repeatedly feeding the model the same text) rises from $0.10 to $0.20 at the standard rate. The Carbon and Argon price lists show no surcharge for long inputs. Companies should check exact amounts for their own usage on the billing screens in Google Cloud or AI Studio.
Source: Gemini API pricing (Google)
Editorial cartoon
