Gemini API Pricing: Free Tier, Paid Models and Hidden Tool Costs
Gemini API pricing combines model tokens, optional service tiers, context caching and tool charges. Google offers a free developer tier, while paid production workloads can choose standard, batch, flex or priority pricing depending on latency and volume needs.
By Aniruddh PanvalkarPricing verified Oct 18, 2026
Plans and pricing
Gemini 3.6 Flash — Standard
US$1.50 / 1M input and US$7.50 / 1M outputFrontier-speed model with search and grounding options for capable production applications.
Best for
Interactive production workloads requiring strong intelligence and low latency.
Not for
Simple bulk processing where Flash-Lite or flex pricing is sufficient.
Gemini 3.1 Flash-Lite — Flex
US$0.125 / 1M text-image-video input and US$0.75 / 1M outputCost-efficient tier for high-volume agentic tasks, translation and straightforward data processing.
Best for
Classification, extraction, translation and high-volume asynchronous workloads.
Not for
Complex reasoning or tasks needing the strongest model quality.
Gemini 2.5 Pro — Standard
US$1.25 / 1M input and US$10 / 1M output up to 200K input tokensStrong multipurpose model for coding and complex reasoning; long prompts use higher rates.
Best for
Coding, long documents and complex reasoning where Pro quality is justified.
Not for
High-volume routine tasks or long prompts without cost controls.
Verified from Google AI for Developers on 2026-07-28. Search grounding, Maps, caching storage and long-context tiers can add charges. Batch processing is commonly 50% below standard rates.
Where Gemini API buyers overpay
The common mistake is comparing only token rates and ignoring service tier, grounding calls and long-context thresholds. Use Flash-Lite or flex for bulk work, standard for interactive traffic, and Pro only where testing shows a material quality gain.
Calculate your exact cost →Who is overpaying for Gemini API?
- Teams paying standard interactive rates for asynchronous batch jobs
- Applications triggering unnecessary search-grounding calls
- Long-context workloads that ignore the higher pricing threshold
Sources
Prices on this page are checked against these vendor pricing pages and catalogue entries.
- ai.google.dev/gemini-api/docs/pricing
- gemini.google.com
- one.google.com/about/google-ai-plans
- gemini.google/subscriptions
- one.google.com/about/ai-premium
- openrouter.ai/google/gemini-2.5-flash
- openrouter.ai/google/gemini-2.5-pro
- openrouter.ai/google/gemini-2.5-flash:batch
- openrouter.ai/google/gemini-2.5-flash-lite
- openrouter.ai/google/gemini-2.5-flash-lite:batch
- openrouter.ai/google/gemini-2.5-pro:batch
- openrouter.ai/google/gemini-2.5-pro-preview-05-06
Frequently asked questions
Does Gemini API have a free tier?
Yes. Google lists free input and output tokens for eligible models with lower limits, while production paid tiers provide higher limits and different data handling.
Is Gemini Batch API cheaper?
Google states that Batch API is designed for asynchronous volume processing and is priced at 50% of interactive standard requests for supported models.