Google Gemini API Pricing & Models (2026 Official Guide)

Google’s current stable general-purpose lineup is led by Gemini 3.6 Flash, Gemini 3.5 Flash, and Gemini 3.5 Flash-Lite. Gemini 3.6 Flash is the latest balance of speed and intelligence, 3.5 Flash targets sustained frontier agentic and coding work, and 3.5 Flash-Lite prioritizes throughput and cost.

All three accept text, images, video, audio, and PDF input and return text. Each supports 1,048,576 input tokens and 65,536 output tokens. The visual comparison above uses the official Gemini Developer API model and pricing pages checked on August 10, 2026.

Choose 3.6 Flash when you want Google’s newest stable general-purpose model and need strong multimodal, grounding, and agentic performance. It also has a lower Standard output price than 3.5 Flash in the current snapshot.

Choose 3.5 Flash for workloads already validated on that model or when sustained frontier coding performance matters. Choose 3.5 Flash-Lite for classification, extraction, translation, routing, and other high-volume tasks where unit cost dominates.

Standard is the normal paid-tier rate. Batch and Flex reduce token prices for work that can accept asynchronous or flexible processing. Priority inference costs more for workloads that need prioritized capacity.

Batch and Flex are not temporary discounts. They are processing options with different delivery characteristics. Check the official documentation for availability and operational limits before routing production traffic.

The three stable models share a broad capability surface: thinking, structured outputs, function calling, code execution, file search, URL context, Search grounding, Google Maps grounding, caching, and preview access to Computer Use.

They do not generate audio or images directly and do not use the Live API. For native image, speech, live conversation, music, or video generation, use Google’s specialized Nano Banana, TTS, Live, Lyria, Imagen, or Veo model families and check their separate units and prices.

The official pricing table lists free-of-charge token usage for these stable models, subject to rate limits and availability. Paid-tier usage is billed at the rates above and is marked as not used to improve Google products, while the free tier is marked as used to improve products.

Free-tier limits are account and service constraints, not a permanent unlimited allowance. Review rate limits and billing status in Google AI Studio before estimating production capacity.

Grounding charges are separate from model token prices. The current Gemini 3.x pricing table includes 5,000 free Google Search requests per month shared across Gemini 3.x models, then $14 per 1,000 requests. Google Maps grounding similarly includes a shared monthly allowance followed by $14 per 1,000 search queries.

One customer request can trigger multiple search queries, and Google charges each individual query performed. Measure actual grounding activity rather than estimating cost only from API request count.

Estimate token cost with: input tokens × input rate plus output tokens (including thinking tokens) × output rate, divided by one million. Add grounding, cache storage, or specialized media charges separately.

This page is a dated editorial snapshot, not a live billing feed. Google can add models, change preview status, or update prices. Verify the model ID, pricing tier, rate limits, and final billing terms on the official Gemini Developer API pages before deployment.