Gemini API pricing in 2026: model costs, service tiers, and when a gateway helps

If you are evaluating gemini api pricing in July 2026, the first problem is not only the price table. It is the fact that search results mix at least four different things: Gemini Developer API pricing, Google AI Studio free-tier usage, Google Cloud or Agent Platform pricing, and consumer Gemini subscriptions.

This guide is narrower and more useful. It uses Google's official Gemini Developer API pricing page and billing docs checked on 2026-07-15, then maps the current numbers to the buying decisions teams actually have to make. It also covers the second decision most search results skip: when gemini api pricing stops being only a Google question because your team also needs GPT, Claude, or other model families under the same billing and routing surface.

In other words, this is both a gemini api pricing explainer and a practical google gemini developer api pricing guide for teams buying production access.

Gemini API Pricing At A Glance

The current official gemini api pricing table is no longer one simple set of rows. Google now splits pricing across current Gemini 3.x models, Gemini 2.5 families, service tiers such as Standard and Batch, and special-purpose surfaces such as Live API audio, image generation, and video generation.

For most buyers, the fastest way to understand gemini api pricing is to start with the token-priced text and multimodal models below.

Model	Standard input $ / MTok	Standard output $ / MTok	Context caching write $ / MTok	Storage $ / MTok / hour	Important notes
Gemini 3.5 Flash	1.50	9.00	0.15	1.00	Free tier exists; Batch cuts token cost in half
Gemini 3.1 Pro Preview	2.00 up to 200k, 4.00 above 200k	12.00 up to 200k, 18.00 above 200k	0.20 up to 200k, 0.40 above 200k	4.50	No free tier; premium Gemini 3 reasoning option
Gemini 3 Flash Preview	0.50 text/image/video, 1.00 audio	3.00	0.05 text/image/video, 0.10 audio	1.00	Shared Gemini 3 grounding allowance
Gemini 3.1 Flash-Lite	0.25 text/image/video, 0.50 audio	1.50	0.025 text/image/video, 0.05 audio	1.00	Lowest-cost Gemini 3 text route in the current table
Gemini 2.5 Pro	1.25 up to 200k, 2.50 above 200k	10.00 up to 200k, 15.00 above 200k	0.125 up to 200k, 0.25 above 200k	4.50	Free tier exists; still relevant for reasoning-heavy work
Gemini 2.5 Flash	0.30 text/image/video, 1.00 audio	2.50	0.03 text/image/video, 0.10 audio	1.00	Supports a 1M token context window
Gemini 2.5 Flash-Lite	0.10 text/image/video, 0.30 audio	0.40	0.01 text/image/video, 0.03 audio	1.00	Cheapest current Gemini 2.5 headline row

That table answers the headline gemini api pricing question, but it still does not tell you what your real bill will look like.

What Changes Real Gemini API Cost

Four pricing modifiers matter more than another generic model roundup:

1. Prompt-size thresholds. Gemini 3.1 Pro Preview and Gemini 2.5 Pro both step up in price once prompts exceed 200k tokens. Long-context teams should treat that threshold as a budgeting breakpoint, not a footnote.
2. Service tiers. Several current models have Standard, Batch, Flex, and Priority pricing. Batch often halves input and output cost. Priority can raise cost materially when you need reserved higher-performance service.
3. Context caching and storage. Google now prices cache writes and storage separately. Reused context can still improve economics, but the storage line item means caching is not free just because it lowers repeated input spend.
4. Grounding and modality mix. Search grounding, Maps grounding, audio token rates, Live API audio, and video output all behave differently from the basic text-token rows.

That is why good gemini api pricing analysis has to separate three decisions:

• Which Gemini model family fits the workload?
• Which service tier fits the latency and throughput requirement?
• Whether direct Google billing is still enough once the team becomes multi-model?

Gemini API Pricing Free Tier Versus Paid Tier

One reason this SERP is messy is that gemini api pricing free tier intent gets mixed into almost every result.

The current official pages make three practical points clear:

• Some current Gemini rows still have a free tier, including Gemini 3.5 Flash, Gemini 3.1 Flash-Lite, Gemini 2.5 Pro, Gemini 2.5 Flash, and Gemini 2.5 Flash-Lite.
• Some important rows are paid-only, including Gemini 3.1 Pro Preview and several special-purpose preview routes.
• Free-tier access is useful for testing and prototyping, but it is not the same buying motion as production budgeting for a steady workload.

That means gemini api pricing free tier should be treated as a testing question, while production gemini api pricing is usually a service-tier and workload-shape question.

Gemini Pro API Pricing And Long-Context Economics

Searchers looking for gemini pro api pricing usually want one of two things: the current premium Gemini rate card, or the answer to whether Pro is worth the jump over Flash-family models.

Today the headline premium rows are:

• Gemini 3.1 Pro Preview at $2.00 input and $12.00 output per million tokens for prompts up to 200k, then $4.00 input and $18.00 output above that threshold.
• Gemini 2.5 Pro at $1.25 input and $10.00 output up to 200k, then $2.50 input and $15.00 output above that threshold.

The practical lesson for gemini pro api pricing is that Pro is not only about higher headline token cost. It is also about whether your prompts cross the 200k threshold often enough that the more expensive long-context tier becomes your actual default.

Gemini API Pricing By Workload

Here is the fastest way to turn gemini api pricing into a deployment decision.

Workload	Best default tier	Why	Watchouts
Classification, extraction, translation, simple automation	Gemini 2.5 Flash-Lite or Gemini 3.1 Flash-Lite	Lowest-cost rows for high-volume traffic	Free-tier testing can hide production throughput needs
General app chat, product copilots, search-grounded assistants	Gemini 2.5 Flash, Gemini 3 Flash Preview, or Gemini 3.5 Flash	Better balance between price and capability for mainstream app traffic	Grounding and audio rates can change the bill faster than text-only estimates suggest
Complex reasoning, coding, agent planning, long-context review	Gemini 2.5 Pro or Gemini 3.1 Pro Preview	Higher-end reasoning and multimodal understanding	Costs jump if prompts regularly exceed 200k tokens
Async backfills and large offline jobs	Same model as above, but use Batch where available	Batch often cuts input and output token cost by 50%	Not suitable for user-facing real-time responses
Live voice or high-quality streaming audio	Live API native audio routes	Better fit for voice UX than standard token-only chat endpoints	gemini live api pricing differs from standard text-model pricing

This is the part most google gemini developer api pricing pages rush past. Teams do not buy a model name. They buy a latency profile, grounding behavior, and unit economics for a real workload.

Gemini API Pricing Math: Three Quick Scenarios

The examples below use official Google list prices only, so the math stays reviewer-safe.

1. High-volume automation on Gemini 2.5 Flash-Lite

Assume 60M text input tokens and 12M output tokens in a month.

Line item	Calculation	Cost
Input	60 x $0.10	6.00
Output	12 x $0.40	4.80
Total		10.80

For this kind of traffic, gemini api pricing is mostly a volume and output-discipline question. Flash-Lite stays attractive when the task is narrow and responses are short.

2. Product assistant on Gemini 3.5 Flash with Batch for offline summaries

Assume 25M input tokens and 6M output tokens.

Line item	Standard pricing	Batch pricing
Input	25 x $1.50 = 37.50	25 x $0.75 = 18.75
Output	6 x $9.00 = 54.00	6 x $4.50 = 27.00
Total	91.50	45.75

This is where gemini api pricing stops being a simple rate-card exercise. Batch can change the economics materially for non-interactive jobs even when the model row itself looks expensive.

3. Reasoning-heavy review flow on Gemini 3.1 Pro Preview

Assume 8M input tokens and 2M output tokens with prompts staying below 200k.

Line item	Calculation	Cost
Input	8 x $2.00	16.00
Output	2 x $12.00	24.00
Total		40.00

For premium reasoning work, the core gemini pro api pricing insight is not that Pro is automatically expensive. It is that cost remains manageable when you keep prompts below the 200k breakpoint and reserve Pro for the tasks that actually need it.

Gemini Live API Pricing Is A Different Surface

One major source of confusion in this SERP is gemini live api pricing.

Live API native audio pricing is not just the standard text-model table with streaming added on top. The current official Gemini 2.5 Flash Native Audio preview row uses a different mix:

• $0.50 input per million text tokens
• $3.00 input per million audio or video tokens
• $2.00 output per million text tokens
• $12.00 output per million audio tokens

That makes gemini live api pricing its own budget category. If you are building voice agents, real-time translation, or streaming assistant flows, do not estimate spend from ordinary Flash chat rows.

Direct Google Access Versus A Gateway

This is where the buying decision changes.

Direct Google access is usually enough when:

• Gemini is the only model family in production.
• Billing can stay inside one Google account and one team workflow.
• The team does not need shared procurement, centralized quota policy, or cross-provider routing.

A gateway starts to matter when:

• You need Gemini plus GPT, Claude, or other providers under one access layer.
• Finance wants one invoice or one prepaid balance instead of scattered provider billing.
• Engineering wants one key and a cleaner route-management surface across providers.
• Ops wants usage review, model-level metering, and shared routing controls in one place.

That is the missing bridge between gemini api pricing and platform evaluation. Once a team is multi-model, the real cost problem becomes operating-model sprawl as much as token spend.

Where Flatkey Fits

Flatkey's current public surfaces support a narrower, review-safe version of that story:

• The homepage currently positions Flatkey around official GPT, Claude, and Gemini access through one key, with headline messaging around prices as low as 50% off official rates.
• The pricing page currently emphasizes prepaid balance, one invoice across providers, usage metering by model and token type, team procurement support, and a shared balance that can route across GPT, Claude, Gemini, image, audio, and video models.
• The live public pricing feed checked on 2026-07-15 returned 170 rows with 95 marked available, including 50 Gemini-family rows total and 28 Gemini-family rows currently marked available.

The safe claim is not "this exact Gemini row is always cheaper by X." The safe claim is that teams evaluating gemini api pricing can centralize access, routing, and spend review once Gemini is no longer the only provider in the stack.

If you want the broader cross-model picture after reading this gemini api pricing guide, compare current access and billing options on Flatkey pricing, then use the AI model pricing comparison page for a wider multi-provider view.

The Real Buying Checklist For Gemini API Pricing

Use this checklist before you lock a model or buying path:

Question	Why it matters
Will prompts often exceed 200k tokens?	Pro-family rows step up in cost above that threshold
Is the workload interactive or async?	Batch can halve cost, but only for non-real-time jobs
Will the app use audio, video, or grounding?	Those surfaces can change real spend much faster than text-only estimates suggest
Is free-tier testing creating false confidence?	gemini api pricing free tier is useful for prototyping, not proof of production economics
Is Gemini the only provider in scope?	If not, gemini api pricing is only one part of the operating-cost story

That framing is what should help this article rank. Searchers want the current gemini api pricing table, but they also want faster decisions after they read it.