Trust page

About nfer

Once you've picked a model, find the cheapest place to deploy it. nfer compares API pricing, dedicated and reserved instances, and rented GPU costs across 25+ providers - for the same model, on the same page.

25+
Providers
100+
Models
30+
Hardware SKUs
600+
Price points

Why it exists#

nfer started life as an internal tool. We were trying to figure out how to reduce our own AI spend and pick the right deployment path for each model we were working with - and we kept hitting the same wall. There's no single place to compare API pricing, dedicated capacity, and rented GPUs side-by-side. We built one for ourselves. It turned out to be the kind of thing other teams could use too, so we opened it up.

There are plenty of tools for picking which model to use. Plenty of great work goes into benchmarking model quality, routing API calls, and indexing the models themselves. Once a team has chosen Llama 3 70B, Mistral, Qwen, or any other model, no comparable tool exists for picking where to deploy it.

Token-priced API providers, dedicated-throughput offerings, and hourly GPU rentals all advertise different prices in different units. nfer normalizes them to monthly cost for your specific workload (your tokens, your I/O ratio, your utilization) and surfaces the cheapest option at a glance, so picking the right deployment is a few clicks instead of an afternoon of spreadsheet wrangling.

What we cover

25+ providers, more than 100 models, 30+ hardware SKUs, and 600+ price points, kept in sync with each provider's published pricing. You can filter by sovereignty, certifications, region, license, quantization, and pricing type. New providers and models are added regularly.

What we're not

A focused tool, not a kitchen sink

Not a quality benchmark

We don't rank models by accuracy or speed - there are plenty of great tools for that. We start where benchmarks end: you've picked a model, now where do you run it?

Not a model recommender

We won't tell you which LLM to use for your task. The choice between Llama 3, Claude, or Mistral is yours. nfer compares deployment economics for whichever model you've already chosen.

Not frontier-only

We over-index on open-source models and the long tail of providers. That's where deployment-target choice actually changes the bill - and where almost no comparator covers properly.

Who's behind it#

nfer is built by M3T, a company building cutting-edge AI products. We're independent: no investors, no provider commissions.

Sizing a workload?

Book a free 30-minute consultation. We can help you choose the right model, select the right infrastructure, support your deployment, and surface the cheapest paths for your specific use case.

Book a free consultation

Popular models

Providers

Learn

About