Google ships Gemini 3.8 Flash TTS and Flash-Lite TTS
- Source
- Artificial Analysis
- Date
Google has released Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS. Gemini 3.8 Flash TTS debuts at #1 on our Pronunciation Robustness Benchmark and #2 on our Provider Voice Arena Leaderboard Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS are @GoogleDeepMind's latest Text to Speech models, supporting preset voices and voice replication. Gemini 3.1 Flash TTS led our Pronunciation Robustness Benchmark at launch, with Gemini 3.8 Flash TTS now taking the top spot and scoring 64 Elo points higher in our Provider Voice Arena. Key takeaways: ➤ #2 in Provider Voice Arena: Gemini 3.8 Flash TTS debuts at #2 with an Elo of 1,263, just behind Sonic 3.6 at 1,272 and ahead of Qwen-Audio-3.0-TTS-Plus at 1,260. Gemini 3.8 Flash-Lite TTS debuts at #6 with an Elo of 1,236, just 1 point behind Simba 3.2. Both models rank ahead of Gemini 3.1 Flash TTS at 1,199. ➤ #1 in Pronunciation Robustness: Gemini 3.8 Flash TTS takes #1 at 89.5%, ahead of Gemini 3.1 Flash TTS at 88.2%, SpaceXAI TTS at 87.6%, and Gemini 3.8 Flash-Lite TTS at 87.4%, placing three Google models in the top four. ➤ Multilingual: Across our 9 multilingual Controlled Voice Arenas, Gemini 3.8 Flash-Lite TTS debuts at #1 in Japanese, #2 in Arabic, and #3 in German. Gemini 3.8 Flash TTS ranks #2 in Japanese and Portuguese. ➤ Speed: Gemini 3.8 Flash TTS processes 44.1 characters per second, compared to 40.2 characters per second for Gemini 3.8 Flash-Lite TTS, approximately 2.7x and 2.4x faster than realtime, respectively. ➤ Pricing: Gemini 3.8 Flash TTS is priced at $32.98/1M characters, compared to $22.07/1M characters for Gemini 3.8 Flash-Lite TTS. Both are more expensive than Gemini 3.1 Flash TTS at $18.31/1M characters, but substantially cheaper than Eleven v3 at $100/1M characters. See below for more details ⬇️
Gemini 3.8 Flash TTS debuts at #2 on the Artificial Analysis Provider Voice Arena Leaderboard, surpassing Qwen-Audio-3.0-TTS-Plus and trailing only Sonic 3.6 Our Provider Voice Arena evaluates Text to Speech models through blind human comparisons using each model's own preset voices. Gemini 3.8 Flash TTS achieves an Elo of 1,263, just 9 points behind Sonic 3.6 at 1,272 and ahead of Qwen-Audio-3.0-TTS-Plus at 1,260. Gemini 3.8 Flash-Lite TTS debuts at #6 with an Elo of 1,236, just 1 point behind Simba 3.2. Both new models rank ahead of Gemini 3.1 Flash TTS at 1,199.

Gemini 3.8 Flash TTS takes the #1 spot on the Artificial Analysis Pronunciation Robustness Benchmark at 89.5%, surpassing Gemini 3.1 Flash TTS at 88.2% and SpaceXAI TTS at 87.6% Our Pronunciation Robustness Benchmark evaluates how accurately Text to Speech models pronounce challenging text across different categories. Key takeaways: ➤ Overall: Gemini 3.8 Flash TTS leads at 89.5%, followed by Gemini 3.1 Flash TTS at 88.2%, SpaceXAI TTS at 87.6%, and Gemini 3.8 Flash-Lite TTS at 87.4%. ➤ Contextually Appropriate: Gemini 3.8 Flash TTS leads at 97.9%. ➤ Expanding Shorthand: Gemini 3.8 Flash TTS leads at 86.1%. ➤ Preserving Exact Sequences: SpaceXAI TTS retains the top spot at 85.7%. ➤ Standalone Terms: Qwen-Audio-3.0-TTS-Plus leads at 95.5%, ahead of v3 Conversational at 95.1% and Gemini 3.1 Flash TTS at 93.7%.

Gemini 3.8 Flash-Lite TTS debuts at #1 in Japanese on the Artificial Analysis Controlled Voice Arena, with Gemini 3.8 Flash TTS taking the #2 spot Our Controlled Voice Arena standardizes speaker identity across models using the same replicated reference voices, allowing us to isolate speech quality from voice preference. We evaluate models across English and 9 additional languages. Key takeaways: ➤ Japanese: Gemini 3.8 Flash-Lite TTS takes #1 with an Elo of 1,212, followed by Gemini 3.8 Flash TTS at #2, giving Google the top two positions. ➤ Arabic and German: Gemini 3.8 Flash-Lite TTS ranks #2 in Arabic with an Elo of 1,274 and #3 in German at 1,250. ➤ Portuguese: Gemini 3.8 Flash TTS takes #2 with an Elo of 1,295. ➤ Other languages: Both models rank in the top 8 in Hindi, Spanish and French. ➤ Gemini 3.8 Flash-Lite TTS and Gemini 3.8 Flash TTS score 1,057 and 1,051 Elo, respectively, trailing Realtime TTS-2 at 1,144.

- In Artificial Analysis' blind Provider Voice Arena, Gemini 3.8 Flash TTS debuts at #2 with an Elo of 1,263, 9 points behind Sonic 3.6 (1,272), and Flash-Lite TTS debuts at #6 (1,236); both beat Gemini 3.1 Flash TTS (1,199).
- Gemini 3.8 Flash TTS leads the Pronunciation Robustness at 89.5%, with Google holding three of the top four spots, but SpaceXAI TTS still leads Preserving Exact Sequences (85.7%) and Qwen-Audio-3.0-TTS-Plus leads Standalone Terms (95.5%).
- The cheaper tier wins in some languages: Flash-Lite TTS ranks #1 in Japanese, #2 in Arabic and #3 in German in the Controlled Voice Arena, which uses the same replicated voices across models to isolate speech quality from voice preference.
- Neither model is fast by TTS standards: Flash TTS generates 44.1 characters per second and Flash-Lite 40.2 (about 2.7x and 2.4x realtime), far behind Falcon 2 at 204.9 characters per second.
- Pricing is $32.98 per 1M characters for Flash TTS and $22.07 for Flash-Lite, both above Gemini 3.1 Flash TTS ($18.31) and SpaceXAI TTS ($15) but well below Eleven v3 ($100).
- benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
Adds two new Google TTS models with measured Elo, speed, and pronunciation scores, giving engineers a concrete data point for choosing a voice provider.
Checking sign-in…
Loading comments…





