{"version":"1.0.0","updatedDate":"2026-10-13","canonical":"https://inquory.com/blog/news/gemini-4-argon-access-pricing","provenance":"Inquory source-based analysis and explicitly labelled synthetic calculations/reference-policy cases. No vendor runtime, account, biology or customer test.","id":"argon","path":"/news/gemini-4-argon-access-pricing","heading":"Reproduce the announced token-cost exposure","contribution":"Three executed token-cost scenarios separating announced introductory and later rates.","method":"Cost = uncached input × input rate + cached input × 5% of input rate + output × output rate, divided by one million. The September 30 announcement supplies $2/$10 introductory input/output rates and $4/$20 later rates. Token counts and cache fractions are Inquory's synthetic assumptions.","paragraphs":["The same short-response example costs 0.400 at the introductory rates without caching, versus 0.229 when 90% of input is cached. Increasing output to one million tokens costs 10.200 with the same uncached input. Output length can dominate token exposure even when input caching is effective.","Amounts below use the announcement's dollar denomination. The general API pricing page labels its own tables in USD, but no Argon row was found on the page opened October 13. Do not treat these announced-rate scenarios as an active account quote, Canadian-dollar conversion or proof of access. Before procurement, obtain an applicable currency, rate, model ID and introductory-period term from the provider."],"headers":["Synthetic workload","Input tokens","Output tokens","Cached input","Intro announced dollars","Later announced dollars"],"rows":[["Short response, no cache","100000","20000","0%","0.400","0.800"],["Short response, 90% input cached","100000","20000","90%","0.229","0.458"],["Long response, no cache","100000","1000000","0%","10.200","20.400"]],"limits":"No API call or billed run. Cache storage, tool calls, retries, thinking-token rules, tax and currency conversion are excluded. Announced rates and a model limit do not establish achievable output, availability, quality or savings.","sources":[{"label":"September 30 Argon announcement, reopened October 13","href":"https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/"},{"label":"API pricing page checked October 13; no Argon row found","href":"https://ai.google.dev/gemini-api/docs/pricing"}],"related":{"label":"Separate token exposure from the full workflow economics","href":"/tools/ai-automation-roi-calculator"}}