LLM API pricing comparison 2026

Compare the live price you pay for every text model available on AnyModel. Filter by provider, sort input or output rates, and calculate the cost of your own token volume before choosing an API.

Live prices are loading — refresh this page in a moment.

AnyModel prices are shown per 1M tokens. Most models use one flat rate; models with separate input/output billing show both rates. Vendor list prices are shown only when directly comparable. No subscription, no minimums.

Image and video pricing

Media is billed per unit of output, not per 1M tokens. A video job costs its duration times the chosen model and resolution rate, up to 15 seconds; an image costs the base rate scaled by quality and size.

WhatUnitBalance tokens≈ price
Grok Imagine Video 480pper second500,000$0.025
Grok Imagine Video 720pper second700,000$0.035
Grok Imagine Video 1.5 480pper second800,000$0.04
Grok Imagine Video 1.5 720pper second1,400,000$0.07
Grok Imagine Video 1.5 1080pper second2,500,000$0.125
Image (gpt-image-2.5)per image, 1024×1024, medium131,700$0.0066

Example: an 8-second 480p clip costs 4M tokens. The charge is booked once, when the provider accepts the job, and polling a running job is free. A job that ends without a video is returned in full — by itself for jobs started in Video Studio, and on the poll that first reports the failure for jobs created through the API.

How to compare LLM API pricing

Input and output rates matter separately. Retrieval, classification and long-document workloads are input-heavy; agents and content generation can spend much more on output.

The AnyModel columns show current customer rates per 1 million tokens. Vendor list prices are context only and may not include caching, batch discounts, regional taxes or special tiers.

Price is only one production variable. Test response quality, latency, rate limits and availability with the same prompts before moving a critical workload.

LLM API pricing FAQ

What does price per 1M tokens mean?

It is the charge for one million input or output tokens. Your estimated cost is input tokens divided by one million times the input rate, plus output tokens divided by one million times the output rate.

Are the prices in this table live?

The AnyModel rates are resolved from the same current pricing catalog used by the service. Model availability and rates can change, so verify the page and dashboard before a production launch.

Is there a monthly API subscription?

No. AnyModel uses a prepaid pay-as-you-go balance with no monthly platform fee or minimum spend.

One key, every model, pay per token

Point any OpenAI-compatible client at AnyModel and switch models by name — no separate billing per provider. Fund one prepaid balance and pay only for what you use.