What is AssemblyAI's LLM Gateway and how does it work?

LLM Gateway is a single API to apply leading LLMs to spoken data. Transcribe audio, then send the transcript and prompt to /v1/chat/completions; Gateway routes to models (OpenAI, Anthropic, Google), preserves transcript structure (speakers, timestamps), and centralizes operations with one integration and unified billing.

Which LLM providers does the LLM Gateway support?

LLM Gateway supports OpenAI (e.g., GPT-5, ChatGPT-4o), Anthropic (Claude family), Google (Gemini 2.5, Gemini 3 Pro), open-source models, and other LLM providersall via one API.

How does LLM Gateway's consolidated billing work across providers?

LLM Gateway centralizes billing for all routed LLMs. Usage from OpenAI, Anthropic, Google, and others is tracked in one AssemblyAI account with unified billing and usage reporting. Charges are per token at model-specific rates shown on the pricing page.

How do I get started with the LLM Gateway?

Create an AssemblyAI account and get your API key. Then POST to https://llm-gateway.assemblyai.com/v1/chat/completions with an Authorization header, a model, and your messages, prompt, or transcript text. LLM Gateway routes to multiple providers (GPT, Claude, Gemini) through one unified API.

What models can I use LLM Gateway with?

LLM Gateway supports 15+ models across Anthropic Claude, OpenAI GPT, and Google Gemini. Examples include Claude 4.5 Sonnet/Haiku, Claude 4 Sonnet/Opus; GPT-5 (Nano/Mini), GPT-4.1, ChatGPT-4o, gpt-oss-20b/120b; and Gemini 2.5 Pro, Flash, Flash-Lite, and Gemini 3 Pro. Select the desired model via the model parameter.

AssemblyAI Cloud

AssemblyAI Voice AI Cloud

Production-ready Voice AI infrastructure

Get industry-leading accuracy at 10x lower cost of ownership than self-hosting, so you can ship Voice AI features without managing GPUs, scaling infrastructure, or stitching together vendors.

Start building View our docs

Model Accuracy & Quality

Low Cost, Low Reliability

Best Value

Poor Value

Expensive and Unscalable

Cost Efficiency at Scale

600+

million inference calls per month

840+

million API calls per month

terabytes of audio processed daily

Build, ship, and scale applications with voice data fast with complete Voice AI infrastructure

Voice AI infrastructure delivering the quality and reliability your applications demand, at economics that scale with your business. Purpose-built for the performance real-world applications require.

One API to ship faster

Ship faster by accessing industry-leading speech models and managed infrastructure through one API integration.

Skip infrastructure management by building on a complete Voice AI platform
Access industry-leading speech models plus managed inference through one API
Redeploy your infrastructure team to building product features that drive revenue
Simplify your Voice AI stack through a single vendor and single support team

Proven reliability and security

Process millions of hours daily with infrastructure built for the reliability, performance, and security production applications require.

Maintain 99.9% uptime SLA with multi-region redundancy and automatic failover
Process millions of hours daily for production applications
Meet SOC 2, HIPAA, GDPR compliance without lifting a finger
Deploy at scale with infrastructure built for enterprise security and performance

Zero infrastructure management

Infrastructure that scales automatically, so you can focus on building features.

Instantly scale from 0 to thousands of concurrent streams
Eliminate GPU provisioning, capacity planning, and hardware optimization
Handle unpredictable voice traffic patterns automatically
Receive infrastructure updates and optimizations with zero effort

Predictable usage-based pricing

Economics that scale with your business.

Pay only for what you use with per-second billing
Eliminate infrastructure overhead including GPU costs, DevOps team costs, and over-provisioning
Reallocate $750K+/year from infrastructure management to product innovation
Avoid minimum commitments and long-term contracts

Quality that matters. Infrastructure that scales.

Why developers choose to build on AssemblyAI's Voice AI infrastructure.

Quality that drives results

Customers consistently report that their end users immediately notice a difference in quality once they switch to AssemblyAI, seeing higher NPS scores, fewer support tickets, and improved customer satisfaction.

Industry-leading WER across accents and domains

Sub-second latency

Purpose-built for real-world audio

Ship faster, stay ahead

Teams building on AssemblyAI report faster ship times and quicker time to value than with any other speech-to-text provider on the market.
‍

Highest customer satisfaction score in the industry

24/7 support with actual engineers

Comprehensive docs and cookbooks

Continuous innovation

With product improvements shipping every week and a track record of setting the bar for speech-to-text quality and accuracy, AssemblyAI evolves ahead of the industry so you can too.

Weekly feature releases

Continuous accuracy improvement

Infrastructure optimized for production applications

Economics that enable growth

Simple, transparent pricing with no upfront commits or contracts, and unlimited concurrency. Our pricing is specifically built to avoid throttles or anything that gets in the way of shipping great products.

Pay-as-you-go pricing with no contracts

Automatic volume discounts as you scale

No engineering overhead costs

Unlock the value of voice data

Build what’s next on the platform powering thousands of the industry’s leading of Voice AI apps.

Try our API for free Contact sales