Desktop DDR5 memory kits

Buying memory comes down to four decisions: how much, how fast, how tight the timings, and how many sticks. The list above answers all four from the makers' own datasheets. What no page here can tell you is what a kit costs today, and the reason for that is a permission problem rather than a technical one.

Define the kit before comparing the price

The list holds 190 desktop DDR5 configurations from 12 makers: ADATA, G.Skill, GeIL, KLEVV, Kingston, Lexar, Mushkin, Patriot, PNY, Silicon Power, TeamGroup and V-Color. Recorded total capacities are 8, 16, 24, 32, 48, 64, 96, and 128 GB, with one, two, or four modules. Exact or manufacturer-stated supported data rates span 4800 MT/s to 8800 MT/s. These are the boundaries of what is listed, not a claim that every capacity, module count, speed, and latency can be combined. Read 26 July 2026: TeamGroup memory catalog.

That last limitation matters. Kingston exposes some capacities and kit sizes alongside supported speed and latency ranges, but its configurator does not prove every Cartesian combination. TeamGroup and Patriot expose exact part-number configurations for many rows. A ranged product-family entry therefore must not be presented as if it were an exact 6000 MT/s CL30 kit unless the cited part number establishes that configuration. Kingston FURY Beast DDR5 and TeamGroup comparison catalog, retrieved 26 July 2026.

Make the choice in this order

1. Memory generation and module type: this list covers consumer desktop DDR5 UDIMMs and matched kits, with some CUDIMM records. It excludes DDR4, laptop SO-DIMMs, server RDIMMs, and ECC-only modules, so it cannot answer those shopping jobs. Kingston FURY Renegade specifications, read 26 July 2026.

2. Total capacity and module count: keep both fields visible. A 32 GB single module and a 32 GB kit of two have the same total capacity but are different recorded products. The matching method also keeps capacity per module and pack count in the identity rather than treating the headline capacity as sufficient. Kingston FURY Beast DDR5, retrieved 26 July 2026 and product-identity methodology, published 2026-07-26.

3. Exact speed and timings: compare offers only after the data rate and latency are known for the exact configuration. The captured exact rows include CAS latencies from CL28 through CL42, while some family-level rows retain a range because the official configurator did not expose one exact combination. TeamGroup memory comparison catalog and Patriot Viper Venom DDR5 page, retrieved 26 July 2026.

4. Profile and part number: check XMP or EXPO support and the manufacturer part number on the exact kit. Profile coverage is not complete across the list, so a missing profile field is unknown rather than proof that the kit lacks it. Kingston FURY Renegade DDR5 and Patriot Viper Venom DDR5 pages, retrieved 26 July 2026.

What the current set can and cannot represent

Thirty-two of the 190 kits carry an official current purchase link on the maker's own page. The other 158 show as US stock not shown, because a maker's catalogue proves the configuration exists without proving a shop has it today. That is a limit of the source, not a judgment about the kit. Read 26 July 2026: TeamGroup T-FORCE catalog and source methodology, published 2026-07-26.

Corsair is absent because its reviewed terms prohibit extraction, and Crucial is absent because its catalog returned a web-application-firewall block. Those omissions mean the 190-entry bounded snapshot is not the entire US DDR5 market. They are recorded gaps, not zero products and not evidence that either manufacturer has no relevant kits. Corsair terms and Crucial catalog, retrieval attempted 26 July 2026 and source registry, reviewed 2026-07-25.

What changes not enough data to judge

A usable RAM observation must preserve manufacturer, part number, capacity per module, module count, data rate, timings, XMP or EXPO profile, condition, merchant, source URL, and fetch time. Ambiguous kit matches do not contribute to coverage or trend. Once the 14-day, 10-observation, and 60% coverage requirements are all satisfied, a current offer can be placed retrospectively within its own stated window. A price-per-gigabyte comparison still needs like-for-like kit identity and a real eligible price; neither is replaced by the catalog count. Product-identity and price-state methodology, published 2026-07-26.