Google Gemini Pricing in 2026
Gemini pricing is confusing. We make it simple. Every model, every tier, every cost in one place with real examples and correct math.
Complete Gemini API Pricing Table
Every current Gemini model with input and output costs per million tokens, context window size, and free tier availability. Prices are for the Google AI Studio / Vertex AI pay-as-you-go tier as of March 2026.
| Model | Input / MTok | Output / MTok | Context | Free Tier |
|---|---|---|---|---|
Gemini 2.5 ProFlagship $2.50 input over 200K context | $1.25 | $10.00 | 1M | 25 req/day |
Gemini 2.5 FlashBest Value $0.30 input over 200K context | $0.15 | $0.60 | 1M | 1,500 req/day |
| Gemini 2.0 Flash Free tier available. Previous-gen workhorse. | $0.10 | $0.40 | 1M | Yes |
| Gemini 1.5 Pro $2.50/$10 over 128K context. 2M context available. | $1.25 | $5.00 | 2M | 25 req/day |
| Gemini 1.5 Flash $0.15/$0.60 over 128K context. Budget option. | $0.075 | $0.30 | 1M | 1,500 req/day |
All prices in USD per million tokens. Context caching is available on all models at 25% of the input price for storage per hour and a 25% discount on cached reads. Prices verified March 2026. Always confirm on Google's official pricing page.
Three Ways to Use Gemini
Google offers Gemini through three distinct channels, each designed for a different audience and use case. Understanding which one fits your needs is the first step to managing costs.
Google AI Studio
The free web-based IDE for prototyping and testing Gemini models. Includes a generous free tier with no credit card required. Ideal for individual developers, students, and early-stage projects.
Vertex AI
Google Cloud's enterprise AI platform for production workloads. Pay-as-you-go pricing with SLAs, VPC networking, data residency controls, and integration with the full Google Cloud ecosystem. Same model pricing as AI Studio.
Gemini Advanced
The consumer subscription at $19.99/mo through Google One AI Premium. Includes the most capable models, Deep Research, custom Gems, Gemini in Gmail/Docs/Sheets, and 2TB of cloud storage. No API access.
Real-World Cost Examples
What does Gemini actually cost for common use cases? Here are three realistic scenarios with the recommended model and correct math.
Chatbot Conversation
Recommended: Gemini 2.5 Flash
A typical chatbot exchange: 1,000 input tokens (user message plus system prompt) and 500 output tokens (assistant reply). At 10,000 conversations per day.
$135/month for 300K conversations
Document Analysis
Recommended: Gemini 2.5 Pro
Analysing a 50-page contract (roughly 50,000 input tokens) with a 2,000-token summary output. Processing 100 documents per day.
$247.50/month for 3,000 documents
Classification Pipeline
Recommended: Gemini 2.0 Flash
Classifying support tickets into categories. 200 input tokens (ticket text plus prompt) and 20 output tokens (label). Running 50,000 classifications per day.
$42/month for 1.5M classifications
Which Gemini Model Should You Use?
Choosing the right model is the single biggest factor in controlling your Gemini costs. The general rule is simple: start with the cheapest model that meets your quality requirements, and only upgrade when you can measure a meaningful difference in output quality.
Use Gemini 2.0 Flash for high-volume, simple tasks
Classification, entity extraction, sentiment analysis, content moderation, and simple Q&A. At $0.10/$0.40 per MTok, this is the workhorse for production pipelines where you need speed and low cost over maximum reasoning depth. For tasks where a correct answer is easily verifiable, 2.0 Flash is almost always sufficient.
Use Gemini 2.5 Flash for the sweet spot of quality and cost
Chatbots, summarisation, translation, content generation, and mid-complexity reasoning. At $0.15/$0.60 per MTok, you get a significant quality improvement over 2.0 Flash for only 50% more on input and the same context window. This is the model most teams should default to for customer-facing applications.
Use Gemini 2.5 Pro for complex reasoning and analysis
Long document analysis, multi-step reasoning, code generation, research synthesis, and tasks requiring deep understanding. At $1.25/$10.00 per MTok, this is 8.3x more expensive than Flash on input. Reserve it for requests where you can demonstrate measurably better output. Many teams use a routing layer that sends 90% of requests to Flash and only 10% to Pro.
Consider Gemini 1.5 models for legacy workloads
Gemini 1.5 Pro ($1.25/$5.00) offers the same input pricing as 2.5 Pro but cheaper output and a 2M context window. Gemini 1.5 Flash ($0.075/$0.30) is the absolute cheapest model in the lineup. These are still supported but Google is focusing development on the 2.x family. Use them if you have tested and validated them for your specific task.
How Does Gemini Compare?
Gemini sits alongside Claude (Anthropic) and ChatGPT (OpenAI) as one of the three major AI platforms. Pricing is competitive across the board, but each has strengths depending on your use case.
Gemini vs Claude Pricing
Gemini 2.5 Pro ($1.25/$10.00) vs Claude Sonnet 4 ($3.00/$15.00) and Gemini 2.5 Flash ($0.15/$0.60) vs Claude Haiku ($0.80/$4.00). Gemini wins on price across the board, with a much larger context window (1M vs 200K). Claude is widely regarded as stronger for coding tasks.
Read full comparison →Gemini vs ChatGPT Pricing
Consumer plans are nearly identical ($19.99 vs $20/mo). For APIs, Gemini 2.5 Flash ($0.15/$0.60) matches GPT-4o mini ($0.15/$0.60), while Gemini 2.5 Pro ($1.25/$10.00) is cheaper on input than GPT-4o ($2.50/$10.00). Gemini's standout advantage is the 1M token context window.
Read full comparison →Save Money on Gemini API Costs
Even with competitive pricing, costs can add up at scale. Two of the most effective strategies are context caching and model routing. Both can reduce your monthly bill significantly with minimal engineering effort.
8 Proven Cost Reduction Strategies
From using the free tier during development to batch processing and prompt optimisation. Eight battle-tested techniques with real savings estimates for each approach.
Context Caching Guide
How context caching works, what it costs, when the break-even point hits, and how it compares to Claude's prompt caching. Essential reading for anyone with repetitive prompts.
Explore Individual Models
Each model has its own pricing nuances, context window behaviour, caching options, and ideal use cases. Dive deeper into the models you are considering.
$1.25/$10.00 per MTok. Full cost breakdown, context caching prices, 5 real-world examples, and comparison with Claude Sonnet 4 and GPT-4o.
$0.15/$0.60 per MTok. High-volume cost modelling, comparison with Claude Haiku and GPT-4o mini, and ideal use cases.
$19.99/mo. Everything included in Google One AI Premium, free vs Advanced comparison, and how it stacks up against ChatGPT Plus and Claude Pro.
Rate limits, daily quotas, what you can build for free, and when you will need to upgrade to paid usage.