- Text Generation
GPT Audio Mini is an OpenAI speech model optimized for low-latency, lightweight audio understanding and generation. It focuses on fast, cost-efficient voice interactions compared with larger…
Powered by Google
Gemini 3.1 Flash TTS Preview
- Text Generation
Gemini 3.1 Flash TTS Preview is Google’s low-latency text‑to‑speech model that generates natural, expressive speech with fine-grained control via style prompts and audio tags. It is optimized for fast, high‑quality voice synthesis across many languages and voices.
Providers Comparison
Speed
Output tokens per second · Higher is better
L
LLM API
S
SiliconFlow (FP8)
G
GMI
C
CoreWeave
N
Novita
D
DeepSeek
P
Parasail (FP8)
D
DeepInfra (FP4)
End-to-End Response Time
Seconds · Lower is better
L
LLM API
S
SiliconFlow (FP8)
G
GMI
C
CoreWeave
N
Novita
D
DeepSeek
P
Parasail (FP8)
D
DeepInfra (FP4)
Price
USD per 1M tokens (blended) · Lower is better
L
LLM API
S
SiliconFlow (FP8)
G
GMI
C
CoreWeave
N
Novita
D
DeepSeek
P
Parasail (FP8)
D
DeepInfra (FP4)
About the model
What is Gemini 3.1 Flash TTS Preview?
Gemini 3.1 Flash TTS Preview is a Google text-to-speech model that converts input text into natural-sounding audio with controllable style and delivery. It is mainly used for real-time voice experiences such as conversational assistants, interactive apps, and accessibility tools that require low-latency, high-quality speech output. It is also suited for content production workflows like audiobooks, podcasts, and voiceovers where expressive, multilingual narration is needed. The model is part of the Gemini 3.1 Flash family and succeeds earlier Gemini Flash TTS variants such as Gemini 2.5 Flash TTS.
Model capabilities
5 Core Capabilities
-
Conversational TTS
Generates natural, conversational speech audio from text prompts, suitable for interactive agents and real-time spoken dialogue applications.
-
Multilingual Speech
Supports speech output in multiple languages and accents, enabling localized voice experiences across diverse global user audiences.
-
Screen Reader Output
Produces clear spoken renderings of on-screen content, assisting in accessibility scenarios like screen readers and reading aids.
-
Image Prompt Narration
Turns model-generated or provided image descriptions into spoken narration, enabling voiceover experiences for visual content pipelines.
-
Text From Images
Reads OCR-extracted text aloud from images or documents, turning recognized visual text into accessible spoken audio output.
Use cases
6 Most Valuable Use Cases
- Real-time Voice Narration
- Audiobook and eBook Reading
- Customer Support Voicebots
- Accessibility Screen Reading
- Interactive Voice Learning
- Voice Output Prototyping
Transparent pricing
Cost Comparison
LLM API offers the lowest TTS costs and latency compared to Gemini Flash TTS equivalents.
| Provider | Region | Latency | Throughput | Uptime | Input ($/1M) | Output ($/1M) | Context |
|---|---|---|---|---|---|---|---|
| LLM API BEST | Global | 80ms | 120 req/s | 99.99% | $0.05/1M chars | $0.05/1M chars | ~30 min audio |
| Global | ~150ms | ~60 req/s | 99.9% | ~$0.60/1M chars | ~$0.60/1M chars | ~30 min audio | |
| OpenAI | Global | ~180ms | ~50 req/s | 99.9% | ~$0.75/1M chars | ~$0.75/1M chars | ~20 min audio |
| Azure AI | US East | ~190ms | ~45 req/s | 99.9% | ~$0.70/1M chars | ~$0.70/1M chars | ~30 min audio |
| Amazon Bedrock | US West | ~200ms | ~40 req/s | 99.9% | ~$0.80/1M chars | ~$0.80/1M chars | ~25 min audio |
Performance benchmarks
Technical Specifications
| Metric | Gemini 3.1 Flash TTS Preview | OpenAI Realtime TTS (gpt-4o-realtime) | Amazon Polly Neural TTS |
|---|---|---|---|
| Avg Latency | ~180ms | ~220ms | ~250ms |
| Max Utterance Duration | ~15min | ~10min | ~5min |
| Price per 1M Characters | $4.00 | $5.00 | $4.00 |
| Languages Supported | ~30 | ~20 | ~30 |
| Voices / Styles | ~40 | ~20 | ~50 |
| Streaming Throughput | ~50 rps | ~40 rps | ~60 rps |
| Avg MOS Quality | ~4.4/5 | ~4.5/5 | ~4.3/5 |
30-day usage via LLM API
- 2.8B
- Input characters synthesized
- 9.4M
- API requests served
- 1.1M
- Unique developer accounts
- 99.8%
- Avg monthly uptime
Architecture & Integration
Why Build on LLM.API?
One unified API. Every major model. Built-in reliability, cost control, and observability.
-
Intelligent Model Routing
Automatically route each request to the optimal model across providers based on latency, cost, and quality so you keep one integration while strategies evolve.
One endpoint, any model. -
Cost-Aware Orchestration
Mix premium and budget models, apply dynamic caps, and analyze per-request spend so you can scale usage without surprise bills or manual tuning.
Optimize tokens, not hacks. -
Resilient Fallback Flows
Define automatic cross-provider fallbacks and retries so outages, rate limits, or timeouts degrade gracefully instead of breaking your production workloads.
Stay up when APIs don’t. -
End-to-End Observability
Trace every call across providers with logs, metrics, and latency breakdowns so you can debug incidents, tune prompts, and prove reliability to stakeholders.
See every token’s journey. -
Task-Level Abstractions
Describe tasks like chat, extraction, or generation once and let LLM.API pick the right tools and models, avoiding provider-specific boilerplate.
Think tasks, not endpoints. -
High-Throughput Batch Jobs
Send large batches across providers with automatic chunking, concurrency control, and retries so you can process millions of items without custom pipelines.
Batch at platform scale.
Decision guide
When to Use — When NOT to Use
Use it if...
- You need fast, low-cost text-to-speech for interactive apps, games, or chatbots.
- Your use case involves prototyping voice features and you can tolerate preview-level stability.
- You need to generate spoken feedback or instructions from short text prompts on-demand.
- Your use case involves turning UI messages or notifications into natural-sounding speech quickly.
- You need a cloud TTS service that integrates easily with other Gemini-family models.
- Your use case involves adding basic voice output to existing web or mobile workflows.
Avoid if...
- You need a fully production-hardened TTS service with strong long-term backward compatibility guarantees.
- Your workload requires strict enterprise compliance certifications or audited data-handling guarantees today.
- You need ultra-low-latency, on-device text-to-speech where cloud round-trips are unacceptable.
- Your workload requires fine-grained control over phonemes, prosody, or custom voice cloning.
- You need guaranteed stable pricing, quotas, and SLAs beyond typical preview-stage offerings.
- Your workload requires multilingual TTS coverage beyond the languages currently supported in preview.
FAQ
Frequently Asked Questions
-
What is Gemini 3.1 Flash TTS Preview?
Gemini 3.1 Flash TTS Preview is a Google model that converts text into speech with a focus on low latency and efficient generation.
-
What is Gemini 3.1 Flash TTS Preview best suited for?
It is best for real-time or near-real-time text-to-speech use cases like voice responses, assistants, and interactive applications where speed matters.
-
Which modalities does Gemini 3.1 Flash TTS Preview support via LLM.API?
Through LLM.API, Gemini 3.1 Flash TTS Preview takes text as input and returns generated audio as output.
-
How fast is Gemini 3.1 Flash TTS Preview in terms of latency?
It is optimized for low latency streaming-style speech generation, making it suitable for responsive conversational experiences.
-
What context window or input length limits apply to Gemini 3.1 Flash TTS Preview?
The model accepts typical TTS-length prompts; very long texts should be chunked by the client before sending for synthesis.
-
How is Gemini 3.1 Flash TTS Preview priced on LLM.API?
Pricing is usage-based on characters or tokens of text-to-speech generation; check your LLM.API dashboard or pricing page for current rates.
-
How do I call Gemini 3.1 Flash TTS Preview through LLM.API?
You select the model name in the LLM.API request and send text input, receiving audio bytes or a URL depending on your integration options.
-
How does Gemini 3.1 Flash TTS Preview compare to non-Flash Gemini models?
Compared to larger, general-purpose Gemini models, it trades broad multimodal reasoning for faster, more efficient text-to-speech generation.
-
What are the main limitations of Gemini 3.1 Flash TTS Preview?
It focuses on speech synthesis only, so it does not perform general reasoning, code generation, or image understanding tasks.
-
Can I fine-tune Gemini 3.1 Flash TTS Preview via LLM.API?
Fine-tuning is not available; you use the base Google-provided TTS voices and control style primarily via prompt parameters.
Compare
Competitive Models
- Text Generation
GLM 4.6 is Z.ai’s flagship mixture-of-experts large language model optimized for coding, reasoning, and agentic workflows. It is notable for its strong performance on code benchmarks…
- Text Generation
Veo 3.1 is Google’s latest high-fidelity video generation model that creates short, cinematic clips from text or image prompts with native audio. It focuses on strong…