Pricing and Free Tier, Fully Explained: $1.50/$7.50, 20 Free Requests a Day, and a 45-Day Notice Period

Hydrogen's price sheet is short: $1.50/$7.50, cached input $0.075, Batch/Flex at half price, free tier $0. But three things about it are easy to misread: it's a standard price, not a promo; the free tier has costs; and the Cloud model page carries a short-term GA clause few people noticed. This article covers pricing, the free tier, and the long-term risk together.

Price table (checked Nov 12–13, 2026; the checkout page is authoritative)

Caching is implicit: prompts of at least 4,096 tokens trigger it automatically, with durable caching on by default; explicit context caching is not supported. Artificial Analysis's blended price (7:2:1 cache hit/input/output) is $1.10 per million tokens.

Standard price vs promo price: don't compare the wrong numbers

Hydrogen's $1.50/$7.50 is a standard price with no expiry date and no announced increase. But the two prices next to it are both promos: Carbon/Argon's $2/$10 expires November 30 and then doubles to $4/$20; 3.8 Flash's $0.75/$3.75 expires December 31 and then rises to $1.50/$7.50.

So today's comparisons are all illusions: Hydrogen looks only 25% cheaper than Carbon/Argon, but 62.5% cheaper after December 1; it looks twice as expensive as Flash, but equal from January 1, 2027. It isn't discounted — the others are. For long-term budgets, use the post–December 1 prices: Hydrogen $1.50/$7.50, Carbon/Argon $4/$20, Flash $1.50/$7.50.

Free tier: the family's only one, with two costs

Hydrogen is the only Gemini 4 model with a free tier — Carbon and Argon are paid-only. The free tier needs no billing account and is good for prototypes; but it has two costs.

The first cost is rate limits. Per scriptbyai's reading of Google's rate-limits page, the free tier allows 5 RPM / 250K TPM / 20 RPD (RPD resets at midnight Pacific; quotas are per project and model). But tbngrow (November 13) says "Google published no numbers on the pages it reviewed" — check Google's rate-limits page for the current figures. Paid-tier limits (scriptbyai): Tier 1: 1K RPM / 2M TPM / 10K RPD; Tier 2: 2K / 3M / 100K; Tier 3: 10K / 20M / unlimited.

The second cost is data use. Free-tier data may be used by Google to improve its products; paid-tier data is not. Production traffic and sensitive content belong on the paid tier.

Cost per task across four levels: the default high is a pricey level

Note: Hydrogen ships at the high level — without changing configuration, you pay $1.87 per task. To save money you have to turn it down yourself: medium at 41 points and $0.80 per task is the value sweet spot for most everyday work. Open xhigh only on tasks where your own tests show a gain — the same budget buys 1.7 runs at high.

The 45-day notice period: what short-term GA means

The Cloud model page lists Hydrogen as short-term GA with a minimum 45-day retirement notice period. Few people noticed this clause, but it matters for long-term planning: it means Google commits to at least 45 days' notice before retiring the model — not that it will retire in 45 days, but that it is not permanent infrastructure. The response: don't hard-code Hydrogen into long-term architecture, abstract the model call behind a layer, keep a migration plan ready, and watch official announcements.