Gemini 4 Argon API pricing doubles on Dec. 1, Google says
Google's Gemini 4 Argon is listed at $2 per 1M input tokens and $10 per 1M output through Nov. 30, then $4 and $20.
Google is charging an introductory rate for Gemini 4 Argon that doubles on a set date. API bills built on the launch price will rise with no change in usage, and the model’s thinking counts toward the output bill.
Google’s Gemini API release notes list Gemini 4 Argon (gemini-4-argon) as generally available on Oct. 13, 2026. The entry calls it “our most capable model built for complex coding, agentic, math, and science tasks.”
Every Paid Rate Doubles on Dec. 1
Google’s pricing page, read Oct. 27, lists standard paid-tier prices of $2.00 per 1M input tokens and $10.00 per 1M output tokens through Nov. 30. From Dec. 1, 2026, the rates are $4.00 and $20.00. Output includes thinking tokens.
Cached input costs $0.10 per 1M tokens through Nov. 30 and $0.20 after. Batch and Flex processing run $1.00 input and $5.00 output, then $2.00 and $10.00. Priority costs $3.60 and $18.00, then $7.20 and $36.00.
Argon is not on the free tier. It supports implicit caching only, and explicit caching is not supported.
Argon Can Spend More of the Output Limit on Thinking
Google’s guide to the model, updated Oct. 14, tells developers with existing apps that set a lower output cap to raise max_output_tokens to at least 32,768. Argon “can spend more of the limit on thinking than earlier models,” the page says. To use fewer tokens, lower the thinking level.
Thinking levels are low, medium and high, with medium as the default. The context window is 1,048,576 tokens, shared by input, thinking and output. The default maximum output for a standard request is 65,536 tokens. A feature Google calls long decode lets a single response run past 64K tokens, up to 1M minus the input, using background execution.
Argon Reads Five Input Types and Writes Text
Google’s model page lists text, image, video, audio and PDF as inputs and text only as output. Computer use is in Preview. The Live API and image generation are not supported.
Argon is available through the Gemini API and Google AI Studio. Agentic video understanding is generally available.
Our View
We think the line that matters most in the price table is the one saying output includes thinking. Google says Argon can spend more of its output limit on thinking and tells developers to raise the cap to at least 32,768 tokens. A higher cap lets each request run longer. Every extra thinking token is charged at the output rate.
At the launch rate, a request that fills a 32,768-token cap costs about 33 cents in output. After Dec. 1 the same request costs about 66 cents. Developers testing Argon now see invoices at half the price the same traffic carries after Nov. 30. The gap grows with every request a coding agent or analysis job sends.
Engineers and the finance leads who sign off on API budgets should price their tests at the Dec. 1 rate. Lowering the thinking level is Google’s stated way to use fewer tokens. Batch or Flex processing, at half the standard rate, cuts the bill for work that can wait.
The lower rate helps only teams that measure output per request now and budget for what that number costs after Nov. 30.