Is Google AI Studio & Gemini API free?

Start with the free website workflow

AI Studio is a developer playground that can turn an original script into speech. This exercise uses the website, with no local model installation. Finish a short script before making a request so the first result is easy to evaluate.

1. Open the speech playground

   Visit Google AI Studio's Generate speech page with an eligible Google account. If access is blocked by account or regional eligibility, resolve it through the official support route before continuing.

2. Keep the test on free access

   Inspect the selected project and any linked key. Choose the Free route, leave billing off and stop if a paid-key or upgrade requirement appears. A subscription elsewhere in Google's products does not establish this project's funding.

3. Select the actual current TTS model

   Look for Gemini 3.8 Flash TTS or Flash-Lite TTS in the available controls. Record the selected model rather than assuming a promotional tile or the default proves access. If only an older model appears, treat that as a different test.

4. Use a preset voice

   Choose an available built-in voice for a short single-speaker exercise. Voice replication and uploaded reference audio add a separate consent and setup workflow that this practice does not require.

Practice: announce a tiny lantern ferry

This original fictional announcement is the first playable sample in the gallery. We used Fola with the Friendly preset: Warm and neutral delivery with micro-pauses. Put the transcript in Speech block text and choose delivery separately through Style. Compare the complete recording with the transcript; a completed request does not prove perfect pronunciation.

• Suggested delivery: a warm, clear public announcement with a brief pause after each sentence. Keep this instruction separate from the spoken words.
• Check lantern ferry and amber lights for rushed syllables. Listen to the final word rather than judging only the opening sentence.
• Save a usable take before changing voice or delivery. Keep the transcript and actual controls together so another listener can understand the test.

What makes this a current-model test?

Google announced Gemini 3.8 Flash TTS and Flash-Lite TTS on September 23, 2026. The current pricing lists Standard Free input and output for both and links to AI Studio. Successful access through the actual browser controls is still a separate check.

Choose the model that is available free for your account and review its result against the script. Newness alone does not establish clearer pronunciation, better acting or a useful download. This page does not repeat Google's quality rankings as our measurements.

Budget requests without inventing a daily audio count

A free price is not unlimited capacity. Google's API limits depend on the model and project, and the active limits can be inspected in AI Studio. The checked pricing does not establish a guaranteed number of complete browser narrations per day.

Start with one short passage and leave room for a correction. A retry can use capacity even when the previous take was unsuitable. If the service reports a quota or capacity limit, keep the script and return later; switching keys does not create a new project allowance.

API daily request quotas reset at midnight Pacific time. Do not present that API schedule as a measured browser refill or convert tokens into a fixed number of polished narration minutes.

Check the downloaded recording

After a successful generation, use the player's Download action and inspect the saved file. Our browser results were mono 24,000 Hz, 16-bit WAV audio. Automation did not capture a file from Download, so we preserved the original WAV already rendered in the page. MP3 listening copies were encoded afterward; no local speech model was used.

Play the complete file outside the generation page. Compare the ending with the transcript, note omitted or extra words and check playback on headphones and a phone speaker. Keep the original file before making a compressed listening copy.

Record the actual model, preset voice, style, date, funding route and measured duration with the transcript. Do not label an unmeasured take watermark-free or promise a universal download quality from one successful file.

Use original scripts and understand free-data handling

The terms describe AI Studio as a developer service for professional or business use. Google does not claim ownership of generated content. This is not a guarantee of exclusivity or clearance for someone else's script, identity or music.

Free interactions may be reviewed and used to improve Google's products. Use a fictional, non-sensitive passage for this exercise. Keep confidential client scripts, personal recordings and private business material out of an unpaid test.

Label shared output as AI-generated and identify the actual service/model. Our reading of the checked terms finds no Free noncommercial-only condition, but input rights, applicable law and Google's use policies still apply.

Keep browser practice separate from an API integration

The existing listing also covers Gemini API developer access. Use AI Studio for hands-on narration practice; a public website need not expose a generation endpoint or ask its visitors for keys to explain that workflow.

An API integration has its own model pricing and project quotas. If you later build one, keep credentials server-side and inspect billing before requests. The free browser test does not establish that every model or feature is free through an API.

When to choose Cartesia instead

Compare another browser voice playground when you want to hear how the same original paragraph is delivered. Cartesia's current pricing lists Sonic-3.6 and 20,000 monthly Free credits.

Cartesia's Free allowance is shared across speech products and commercial licensing starts on a paid plan. Its account controls and voice quality have not been tested here; the allowances are not interchangeable.