Gemini API Free Tier in 2026: What's Free, What Isn't, and When You Pay

What the Gemini API Free tier includes

Google describes the Free tier as "For developers and small projects getting started with the Gemini API." It includes limited access to certain models, free input and output tokens, and Google AI Studio access (checked Oct 20, 2026). See the official Gemini API pricing page.

The pricing page lists free input and output for Gemini 3.8 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 3.8 Flash TTS, Gemma 4, and Gemini Embedding 2 (checked Oct 20, 2026). Confirm the current list on the official pricing page.

The same page marks Gemini 4 Argon, Gemini 3.1 Pro Preview, image models such as Gemini 3.1 Flash Image, Gemini 3 Pro Image and Nano Banana 2.1, Veo 3.1, and Lyria as unavailable on the Free tier (checked Oct 20, 2026). Check the official model table before choosing a model.

Free tier vs paid: what changes

On the Free tier, content is marked as used to improve Google's products, while paid-tier content is marked as not used; Google Search and Maps grounding is unavailable on the Free tier for most models (checked Oct 20, 2026). Review the official pricing comparison.

Google does not publish per-model Free-tier RPM, TPM, and RPD values on the rate-limits page; view them in Google AI Studio. Limits apply per project rather than per API key, and daily quotas reset at midnight Pacific time (checked Oct 20, 2026). The paid thresholds shown are Tier 1 with linked billing, Tier 2 after $100 cumulative spend and 3 days, and Tier 3 after $1,000 and 30 days. See Gemini API rate limits.

Upgrading means linking a billing account and prepaying at least $5; returning to Free requires disabling billing on each project. Gemini API usage is excluded from the $300 Google Cloud Free Trial, and accounts opened after March 2, 2026 cannot use Welcome credits for Gemini API or AI Studio (checked Oct 20, 2026). See the official billing guide.

When to switch to paid

Switch when a required model or grounding feature is unavailable, your project reaches its current Studio limit, or production needs make paid data handling and predictable capacity more appropriate. The table below uses 800 input and 300 output tokens per request, no caching, retries, tools, or batch discount.

These are planning estimates for 1,000, 10,000, and 100,000 requests per month. Your invoice can differ when requests use more tokens, retries, caching, long-context tiers, tools, or other billable features.

Compare legitimate free and low-cost starting points

If Gemini does not fit the project, compare Claude API starter credits, the OpenAI Free usage tier and paid path, and the monthly ranking of the cheapest LLM APIs for side projects. Each option has different access, data, model, and billing constraints.

FAQ