Gemini 4 Argon pricing
Google published Gemini 4 Argon's token pricing in the September 30 launch announcement, so there is nothing to estimate here — the numbers below are the official ones, plus what they mean for your bill once the model actually reaches your account.
Short answer: $2 per million input tokens and $10 per million output tokens during the introductory period. After it expires: $4 per million input and $20 per million output. Cached input tokens are 95% off the input price.
Gemini 4 Argon official token pricing
| What you pay for | Introductory price | After the introductory period |
|---|---|---|
| Input tokens | $2.00 / 1M | $4.00 / 1M |
| Output tokens (includes thinking tokens) | $10.00 / 1M | $20.00 / 1M |
| Cached input tokens | 95% off input — about $0.10 / 1M* | 95% off input — about $0.20 / 1M* |
What Google has not published
- No end date for the introductory period. Google says only that the $4/$20 rates apply "after the introductory period expires." No date, no token threshold.
- No Batch, Flex or Priority tier for Argon. Those discounted tiers exist for the Flash models. Nothing equivalent has been announced for Argon, so do not assume the usual 50% batch reduction applies.
- No free tier. No free-of-charge allocation has been announced for Argon.
- No regional pricing. The published figures are USD.
Can you pay for it yet?
Not yet. This is the part most coverage leaves out. Argon is rolling out to trusted cyber defenders through Google's Fairwind Program first, with Google AI Ultra subscribers and paid API customers named as the next group. The model is not yet listed on Google's own Gemini API pricing page, which is the clearest evidence that it has not been switched on for developers.
| Audience | Status |
|---|---|
| Fairwind Program cyber defenders | rolling out |
| Google internal teams | in use |
| Google AI Ultra subscribers | announced, not yet live |
| Paid API customers | announced, not yet live |
| Enterprise, developers, consumers generally | no timeline given |
Practical consequence: the introductory price clock has not started for most people. If the introductory window is what makes Argon affordable for your workload, the moment to watch is when it appears on the API pricing page — that is the signal that billing has begun.
How the cached input discount works in practice
Context caching is the single biggest lever on an Argon bill, because it applies a 95% discount to the input side rather than the output side. Two things follow from that:
- Long, stable prompts and large documents benefit most. If the same large context is sent on every request, almost all of it can be cached, and the input portion of the bill drops to roughly a twentieth of the standard rate.
- Chatty, fast-changing prompts benefit least. Text that changes every turn cannot be reused from cache, so the discount does not apply to it.
Because Argon's output is priced at five times its input rate in both periods, workloads that generate long outputs — code, long-form drafting, multi-step agent traces — are dominated by the output line, not by caching.
Gemini 4 Argon vs the rest of the Gemini line
Argon is the most expensive model in Google's current lineup, and it is the only frontier-tier model priced above the Flash family. The comparison below uses Google's published rates for the Flash models, which run on promotional pricing through December 31, 2026.
| Model | Input / 1M | Output / 1M |
|---|---|---|
| Gemini 4 Argon (introductory) | $2.00 | $10.00 |
| Gemini 4 Argon (standard) | $4.00 | $20.00 |
| Gemini 3.8 Flash | $0.75 | $3.75 |
| Gemini 3.7 Flash | $0.75 | $3.75 |
| Gemini 3.6 Flash | $0.75 | $3.75 |
| Gemini 3.5 Flash | $1.50 | $9.00 |
The pricing tells you what the model is for. At roughly 2.7× the input and 2.7× the output rate of 3.8 Flash on introductory pricing — and 5.3× once standard pricing kicks in — Argon is not meant for high-volume, low-complexity calls. It is priced for long-horizon work: multi-hour coding tasks, large codebase migrations, document-heavy research. Google's own write-up leans on exactly those examples, including a 1M-token output limit (up from 64K) that only makes sense if you intend to let the model generate for a very long time.
What a bill actually looks like
Two illustrative workloads, priced at the introductory and standard rates. Neither includes caching, which would lower the input figures substantially.
| Workload | Introductory | Standard |
|---|---|---|
| 1M input + 1M output tokens | $12.00 | $24.00 |
| 10M input + 2M output tokens | $40.00 | $80.00 |
Gemini 4 Argon pricing FAQ
How much does Gemini 4 Argon cost?
An introductory $2 per million input tokens and $10 per million output tokens, then $4 per million input and $20 per million output after the introductory period. Cached input tokens are 95% off the input price.
Is there a free tier for Gemini 4 Argon?
No free tier has been announced. As of October 1, 2026 the model is not listed on Google's Gemini API pricing page at all, because it has not yet been opened to developers or paid API customers.
When does the introductory price end?
Google has not published an end date. The announcement only states that standard pricing applies after the introductory period expires.
What does the cached input discount mean in dollars?
95% off input tokens — about $0.10 per million at the introductory rate and about $0.20 at the standard rate. Those dollar figures are our arithmetic from Google's stated percentage, not numbers Google published separately.
How does it compare with Gemini 3.8 Flash?
Argon costs more per token than every Flash model currently offered: $2/$10 introductory against Flash's $0.75/$3.75 promotional rate, and $4/$20 standard afterwards. Argon is the frontier tier, not the volume tier.
Sources
- Gemini 4 Argon: our next era of frontier intelligence — Google, September 30, 2026 (primary source for the introductory rates and the cached-input discount)
- Google announces Gemini 4 Argon as its new frontier model — 9to5Google, September 30, 2026 (source for the post-introductory $4/$20 rates and the rollout order)
- Gemini Developer API pricing — Google (click-through rate reference for the Flash models; note that Argon does not yet appear here)