Pricing

What the marketis actually doing.

Real secondary-market ranges, computed from observed data, not pulled from thin air. What you see is what the market is doing.

Market rates

Secondary market pricing for enterprise GPUs

Ranges reflect current availability and condition across the active catalog.

live market
Updated June 6, 2026/ researched ranges, live feed coming

Generation bands

Hopper

2 models

$22,000to$35,000

Ada

4 models

$2,300to$8,000

Ampere

6 models

$700to$15,000

$700$35,000

12

Models priced

53

Price observations

20

Vendor sources

H100 PCIe 80GB

H100 PCIe

Hopper

80GB / PCIe

Drop-in Hopper performance for standard server chassis.

Best for / Flexible inference + training

Market range

$22,000to$35,000

H100 NVL 94GB

H100 NVL

Hopper

94GB / NVL

Paired NVL cards with extra memory for big-model inference.

Best for / Large-context LLM serving

Market range

$28,000to$35,000

A100 PCIe 80GB

A100 PCIe

Ampere

80GB / PCIe

The proven Ampere workhorse, abundant in the secondary channel.

Best for / Cost-efficient training

Market range

$5,000to$15,000

RTX A6000 48GB

RTX A6000

Ampere

48GB / PCIe

Ampere workstation card with 48GB for rendering and inference.

Best for / Pro viz + inference

Market range

$3,500to$4,800

L40S 48GB

L40S

Ada

48GB / PCIe

Versatile Ada accelerator for inference and visual compute.

Best for / Inference + graphics

Market range

$5,200to$8,000

A40 48GB

A40

Ampere

48GB / PCIe

Reliable 48GB Ampere card for general-purpose acceleration.

Best for / Mixed workloads

Market range

$4,700to$7,500

A100 PCIe 40GB

A100 PCIe

Ampere

40GB / PCIe

The 40GB Ampere workhorse, abundant and affordable in the secondary channel.

Best for / Cost-efficient training

Market range

$5,000to$8,000

RTX 6000 Ada 48GB

RTX 6000 Ada

Ada

48GB / PCIe

Ada-generation workstation card with 48GB for rendering and inference.

Best for / Pro viz + inference

Market range

$5,200to$7,500

L4 24GB

L4

Ada

24GB / PCIe

Low-power Ada card built for high-density, energy-efficient inference.

Best for / Efficient inference

Market range

$3,000to$3,800

A30 24GB

A30

Ampere

24GB / PCIe

Compact Ampere accelerator for cost-sensitive inference workloads.

Best for / Entry inference

Market range

$1,900to$4,500

RTX 4090 24GB

RTX 4090

Ada

24GB / PCIe

Flagship Ada consumer card, popular for development and small-scale training.

Best for / Dev + small-scale

Market range

$2,300to$2,900

RTX 3090 24GB

RTX 3090

Ampere

24GB / PCIe

Previous-gen consumer flagship, a budget entry into 24GB compute.

Best for / Budget experimentation

Market range

$700to$1,050

The model

What drives GPU pricing

Six factors set the number. We weigh all of them against live data.

01 /

Model & Generation

Hopper commands a premium over Ampere; architecture sets the floor.

02 /

Memory Configuration

More on-card memory widens the workloads a buyer can run, and the price.

03 /

Form Factor

SXM, PCIe, and NVL carry different interconnect value and chassis fit.

04 /

Condition

New, refurbished, and used each price differently against documented provenance.

05 /

Quantity

Volume moves the number in both directions depending on supply depth.

06 /

Warranty & Documentation

Testing records, certifications, and warranty coverage de-risk and lift value.

Get started

Need a number you can trust?

Tell us what you need or what you have. We handle the rest, from first conversation to final payment.