Llama 3.3 70B Azure Pricing

How token costs are calculated

For each region and deployment type, multiply each token category by its matching rate, then divide the sum by 1,000,000. Monthly estimates multiply that request cost by the number of requests. Ranges use complete workloads priced in each region, so input and output prices from different regions are never combined.

Request cost = (uncached input tokens × input rate + cached input tokens × cached input rate + cache-write input tokens × cache-write rate + output tokens × output rate) / 1,000,000.

The examples below use zero cache writes. The cached example assumes half of the input tokens are eligible cache reads on every request; actual cache hit rates and any cache creation charges depend on the deployment. Missing cache rates are not replaced with ordinary input rates. Estimates cover these token meters only, before taxes, negotiated discounts and any separate hosting, tools, media or provisioned-capacity charges.

Some model token limits are not published in the metadata. These are arithmetic cost examples; verify that your deployment supports the workload before using it.

Model information and limits

Open Llama instruction model for multilingual chat, reasoning, and coding

Metadata model ID

llama-3.3-70b-instruct

Metadata match

model

Input modalities

text

Output modalities

text

Context window (tokens)

128,000

Maximum input tokens

Not published

Maximum output tokens

32,768

Reasoning

No

Tool calling

Yes

Structured output

Not published

Supplemental metadata comes from Models.dev, with verified Azure version corrections where available. Family matches do not establish version-specific limits. Confirm capabilities and lifecycle in the official model catalog.

Price sources and deployment guidance

Rates come from the Azure Retail Prices API. This page covers consumption token meters in Azure Public. Provisioned throughput, training, non-token media meters and sovereign-cloud pricing are separate billing scopes.

• Microsoft Foundry pricing

  Review the official platform pricing overview and Azure purchasing guidance.

• Microsoft Foundry model catalog

  Browse the broader official model catalog, providers, APIs, and capabilities.

• Foundry Models sold by Azure

  Check model IDs, context windows, APIs, capabilities, and lifecycle guidance.

• Azure OpenAI pricing

  Review Microsoft's public model tables and current Azure OpenAI list prices.

• Foundry deployment types

  Understand data processing scope, pay-per-token, Batch, and provisioned PTUs.

• Azure Retail Prices API

  See the Microsoft API documentation for the retail meter data used here.