Skip to content

Models

The models your providers serve.

293 models across 32 provider kinds. INFRO routes each request to a model through a provider account you connect, so what you can call is what your connections serve.

Rates are each vendor's published list price, reconciled 2026-10-05 — what your provider bills you, not an INFRO charge. An organization on a negotiated rate enters its own in the console. INFRO adds 0% to model spend; the platform is a subscription. See every provider you can connect. Moving from another gateway or router? Read the migration guides.

Filters

293 models

Browse by vendor

Every model, grouped by the company that makes it. Each vendor page lists its models with list rates and the providers that serve them.

Alibaba models 25

Amazon models 4

Anthropic models 12

AssemblyAI models 2

Black Forest Labs models 7

Bria models 2

ByteDance models 15

Canopy Labs models 3

Cartesia models 1

Clarity AI models 1

Cloudflare models 3

Cohere models 4

Deepgram models 4

DeepSeek models 3

ElevenLabs models 10

fal models 1

Fireworks models 1

Fish Audio models 5

Google models 23

HeyGen models 1

IC-Light models 1

Ideogram models 3

Inception models 1

Kokoro models 1

Kuaishou models 8

Liquid AI models 1

Luma models 3

Magnific models 2

Meta models 3

Microsoft models 7

MiniMax models 15

Mistral models 10

MMAudio models 1

Moonshot models 4

NVIDIA models 4

OpenAI models 37

Perplexity models 5

PixVerse models 2

Qwen models 2

Real-ESRGAN models 1

Recraft models 7

Runway models 4

Sesame models 1

Stability AI models 1

Suno models 1

Sync models 4

Thinking Machines models 1

Topaz Labs models 2

TypeSafe models 1

Upstage models 2

xAI models 16

Z.ai models 15

Questions about this model reference

Where do these model rates come from?
Each rate is the model vendor's own published list price, read from the vendor's pricing page and reconciled 2026-10-05. Your provider account bills you at the rate on that account, so a negotiated rate replaces the list rate.
Does INFRO charge for model usage?
No. INFRO is a BYOK control plane: you connect your own provider accounts and they bill you for inference. INFRO charges a software subscription for routing, failover, observability, budgets and policy, and adds 0% to model spend.
What does the provider count on a model mean?
How many of the provider kinds INFRO connects can serve that model, by the gateway's own routing rule. You call a model through INFRO once you connect an account with one of them.
Can I call these models with the OpenAI SDK?
Yes. Point the OpenAI SDK at https://api.infro.io/v1 with an INFRO gateway key and pass the model id shown on each page. Images, video and audio use their own endpoints, listed on each model page.