GEMINI TTS PRODUCTION COMPARISON WORKSHEET
Checked October 10, 2026 | Burki free resource
Guide: https://burki.dev/blog/gemini-flash-tts-vs-flash-lite-tts

This is an unperformed evaluation protocol, not benchmark results.
Free worksheet does not mean free production voice usage.
Neither Gemini TTS model is currently exposed in Burki's voice-provider selector.
No paid requests, clone creation or account changes are authorized by this worksheet.

1. FROZEN PRODUCTION CONTRACT
Owner / reviewer / planned authorized budget:
Dataset version / source permission / retention:
API path: Interactions or GenerateContent (record the actual chosen path)
SDK version / region / access date:
Exact pair: gemini-3.8-flash-tts / gemini-3.8-flash-lite-tts
Permitted voice identifier, identical for both:
Language / target locale / intended listener group:
Delivery metadata, identical for both:
Unary or streaming / encoding / sample rate / channels / playback application:
No cloning or persona change in the baseline.

2. SHORT SCRIPT
Your support request is waiting for staff review.
Please keep the device identifier and a callback number ready.
This message does not confirm a repair visit.

Criteria: all words audible; review status clear; no lost final qualification.
Listener: What, if anything, has been confirmed?
Actual answer / omitted or changed words:

3. THREE-TURN SCRIPT
Turn 1:
Before you describe the fault, find the identifier printed on the device.
If a character is unclear, say that you are unsure rather than guessing.

Turn 2:
Next, explain what happened and when you first noticed it.
A short sequence of events is more useful than a proposed diagnosis.

Turn 3:
Finally, give the team a callback number and your preferred time.
Staff will review the information. A service visit has not been arranged.

Keep one speaker in the baseline. Review continuity, volume, turn boundaries and qualifications.
Optional later two-speaker variant: record explicit speaker assignment for every turn.
These scripts are English only and fictional. Do not treat them as a language benchmark.

4. PER ATTEMPT RECORD (copy for every attempt, including failures)
Blinded label / actual model (reveal after listening):
Request / input / voice / style / format version:
Actual API result / error / retry reference:
Time request sent:
Time first playable audio byte received:
Time complete output received:
Time actual playback ended, if measured:
Generated audio duration:
Input text usage / output audio usage / billing meter:
Words changed or omitted:
Qualification retained? Evidence:
Voice continuity / volume / transition observation:
Listener preference and reason (separate from text correctness):
Accepted production output? Why:

5. COMPARABLE SUMMARY
Attempts per model / successes / failures / retries:
Same script, voice, format, region and timing boundaries? Exceptions:
Metric / sample count / aggregation / conditions:
Record request-to-first-byte, completion and playout separately.
No common exact-pair benchmark or winner is supplied by Burki.
Do not assign scores from another Gemini model without verified identity mapping.

6. COST WORKSHEET
Standard Gemini Developer API, USD per million tokens, checked October 10:
Through December 31, 2026: text input 0.50 both; audio output Flash 9.00, Lite 6.00.
From January 1, 2027: text input 1.00 both; audio output Flash 18.00, Lite 12.00.
Provider conversion: 25 output audio tokens per second.

Illustration only: 1 hour output = 90,000 audio tokens.
Current output-only cost: Flash USD 0.81 / Lite USD 0.54.
From January 1: Flash USD 1.62 / Lite USD 1.08.
Excludes text input, retries, application, storage, telephony and platform charges.
Actual account terms / price checked date:
All attempts input cost:
All attempts output cost:
Retries / other services:
Accepted output duration:
Total actual cost per accepted output minute:
Keep Standard separate from Batch, Flex, Priority and free-tier conditions.

7. BURKI CONFIGURATION BOUNDARY
Current available provider / model / permitted voice, if evaluating Burki separately:
Readiness / Usage & billing checked:
Actual authorized configuration and execution evidence:
Do not paste Gemini IDs into another provider's custom voice field.
Current selector availability is not provider acceptance or caller-heard verification.

PRIMARY REFERENCES
https://ai.google.dev/gemini-api/docs/changelog
https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash-tts
https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash-lite-tts
https://ai.google.dev/gemini-api/docs/speech-generation
https://ai.google.dev/gemini-api/docs/pricing