Monthly cost estimate (rough)

Estimates paid-tier costs for AI Keyboard’s two Gemini features: live transcription and output style rewriting. Free tier is often $0 but can be quota-limited.

This is a rough calculator (not a bill). Actual cost is based on the token usage reported by the Gemini API.

Models used by AI Keyboard

Gemini Live (streaming STT): gemini-2.5-flash-native-audio-preview-12-2025
Output styles (Fix Grammar / Professional / Casual): gemini-2.5-flash-lite

Tip: If you want the cheapest option, use Android transcription and only use Gemini for output styles.

Estimator

Pricing sources: Google pricing + tokenization docs

API billing tier

Free (often $0) Paid (estimate cost)

Free tier pricing is listed as “Free of charge”, but usage limits can apply.

Transcription provider

Android (Free) Gemini Live

Live transcription costs are dominated by audio input tokens (32 audio tokens/sec).

Minutes of speech per day 10 min/day

Typical daily use: 5–15 minutes.

Days per month

Speaking speed (words/min)

Conversation is often around ~150 wpm (varies).

Use Gemini output styles 100%

Set to 0% if you only want free transcription and never run AI rewriting.

Rewrite length (output vs input) 1.0 x

Most rewrites are similar length.

Estimated monthly total

$—

Range reflects tokenization uncertainty.

Gemini Live transcription

$—

Audio input + transcript output.

Gemini output styles (Flash-Lite)

$—

Text in + rewritten text out.

Estimated tokens

Speech transcript size (text): —
Gemini Live audio input: —
Gemini Live text output (transcript): —
Style input (text): —
Style output (text): —

Quick examples

Same assumptions, different daily speech time. “Android + styles” is usually near-zero cost.

Minutes/day	Android + styles	Gemini Live + styles

References

Pricing: Gemini 2.5 Flash Native Audio (Live API) · Gemini 2.5 Flash‑Lite
Tokenization: audio is 32 tokens/sec (Gemini token docs)
Speaking rate: average conversation ~150 wpm (reference)