How to read a CPU model name: Core Ultra and Core suffixes, and why the same model performs differently

Who this guide is for and what to prepare

• Readers who want to know what the digits and the trailing letter in a laptop's CPU name actually mean
• Readers puzzled that reviews of two machines with the same CPU name reach different conclusions
• Desktop builders who want to avoid the classic mistake of buying a suffix that needs a separate graphics card

What you need

• Copy the CPU name from each candidate's spec sheet exactly, including the trailing letters
• Check whether the same spec sheet publishes a power figure at all (Base Power, TDP, maximum power draw)
• If you are building a desktop, decide in advance whether you will fit a separate graphics card

1. Split the name into tier, generation and suffix

A CPU name contains a brand and tier portion, digits that carry the generation or a sequence number, and a trailing letter that signals the design direction. Where the tier ends and the generation begins differs between manufacturers, so applying one vendor's reading rules to another vendor's name produces wrong answers.

Intel's official processor numbers page lists the Core Ultra brand tiers as "Intel Core Ultra 9 processor, Intel Core Ultra 7 processor, and Intel Core Ultra 5 processor" and the Core brand tiers as "Intel Core 7 processor, Intel Core 5 processor, and Intel Core 3 processor" (read 2026-09-20).

On generations, the same page states that "Intel Core Ultra processors have a SKU number 1 or 2, representing the generation, respectively." So on the observation date, 1 and 2 were the numbers the page described as generation markers. More may be added when the page is updated, so this guide does not claim those two are the only ones there will ever be. The page continues: "Intel Core Ultra mobile processors have suffixes H, U, or V" and "Intel Core Ultra desktop processors have suffixes K, F, KF, or T".

2. The suffix list, in Intel's own words

The table below collects the suffixes shown on Intel's processor numbers page together with the definition each one is given there. The definitions are quoted from the page; they are not this guide's paraphrase.

The suffix that causes the most expensive mistakes is desktop F, which Intel defines as "Requires discrete graphics". Build a desktop around an F part without a separate graphics card and nothing appears on the monitor. KF has no row of its own in the suffix table, but the body text names the combination: "Intel Core Ultra desktop processors have suffixes K, F, KF, or T".

Two caveats. The same table also lists embedded suffixes (E, UE, HE, UL, HL), which are omitted here because they do not arise in an ordinary purchase. And V: the body text names H, U and V as the Core Ultra mobile suffixes, but no V row appears in the suffix table. Its existence is stated; a definition is not given there. This guide does not supply a meaning for a suffix the official page leaves undefined, so look a V model up on the manufacturer's own product page.

3. Why the same model number is not the same machine: two power fields

The claim that identical CPUs behave differently in different laptops has a manufacturer-published basis. AMD's official processor specifications database carries Default TDP and AMD Configurable TDP (cTDP) as two separate fields, the second being the range of power budgets a device maker may set.

The table below quotes values straight out of that database (the page was fetched and read on 2026-09-20). The Ryzen AI 7 350 has a Default TDP of 28 W but a cTDP of 15-54 W. The same model number therefore covers a thin machine set to 15 W and a thicker one set to 54 W, a factor of more than three in the budget the maker is allowed to give it.

Note what this does not say. It is not a claim that X watts yields Y per cent more speed. This site runs no benchmarks and will not put a number on the relationship between the power budget and measured performance. The supported fact is narrower and still useful: the budget itself is not fixed by the model number.

4. Four fields to read next to the model number

AMD's database carries more than the power fields. As the table above already shows, the integrated Graphics Model can differ between parts that sit one step apart in the tier list. The Ryzen AI 5 340 and the Ryzen AI 5 330 differ by a single digit in the name but carry Radeon 840M and Radeon 820M respectively.

A close tier number therefore does not imply the same integrated graphics. The number of display outputs and whether a light game runs at all can change here. If you are not sure which graphics your display is currently attached to, this site's guide on integrated versus discrete graphics covers the check.

The table below lists the fields worth reading on the same page once you have found the model. Each of them is something you can read as a published value rather than infer from the name, which makes them a fair common axis when you line candidates up.

5. The check to run before you buy

Four steps. First, copy the CPU name from the spec sheet exactly, trailing letters included. Second, if it is an Intel part, look the suffix up on the processor numbers page and confirm it is not desktop F. Third, if it is an AMD part, search the model in the official specifications database and note Default TDP, cTDP and Graphics Model.

Fourth, look at whether the machine's own spec sheet publishes a power figure. If it does, you can compare that value between candidates. If it does not, you are deciding on the understanding that the machine may not behave like another machine carrying the same CPU name. Knowing precisely what you do not know is the practical output of this step.

With two or more candidates, write those four items out in the same shape before comparing. A comparison table that lists only CPU names hides exactly the difference this guide is about.

6. What this guide deliberately does not cover

It does not decode AMD's four-digit mobile numbers digit by digit. AMD's own article explaining that scheme now redirects to AMD's blog index, and the text could not be read on the observation date. Without a readable primary source this site will not assert what each digit means. Look the individual model up in the specifications database instead.

It does not compare performance between model numbers. This site runs no benchmarks and does not reproduce other people's measurements as a performance ratio. Prices, discount timing and stock are outside its scope for the same reason.

One access note, since it affects reproducibility. Intel's processor numbers page answers differently depending on the client: a command-line HTTP client received HTTP 403, while another client retrieved both the English and Japanese editions with HTTP 200. Everything quoted above was read from those successful retrievals.

Limitations and requirements

• A suffix records the direction a manufacturer designed the part for. It does not guarantee measured performance or acoustics. "An H part is always fast" and "a U part is always quiet" are both unsupported readings.
• Intel defines the desktop F suffix as "Requires discrete graphics". Build with one and no separate graphics card and you will get no picture.
• The tier digit (9 / 7 / 5) and the generation digit are different axes. Asking whether Core i7 or Core Ultra 7 is higher mixes brand and generation, and the model numbers alone cannot answer it.
• AMD's own article explaining the meaning of each digit in its four-digit mobile numbers could not be reached on 2026-09-20. Do not settle the question from second-hand summaries.

Frequently asked questions

Sources and verification date

Sources checked: 2026-09-20. These sources support the specifications, procedures or prices discussed here. Check each source for applicable conditions and current information.

• Intel: Processor Numbers (brand tiers, the statement that a Core Ultra SKU number 1 or 2 represents the generation, the body text listing the Core Ultra mobile and desktop suffixes, and the suffix table. English edition, retrieved with HTTP 200 on 2026-09-20) ↗
• Intel: Processor Numbers, Japanese edition of the same page. The Japanese wording used in the Japanese edition of this guide is taken from here. Read 2026-09-20 ↗
• AMD: processor specifications database (that Default TDP and AMD Configurable TDP (cTDP) exist as separate fields, and the per-model values quoted in the table above. Read 2026-09-20) ↗

Related practical guides

• CPU value in Japan, September 2026: five gaming candidates ranked
• Intel B860 vs. Z890 vs. H810: Choose an LGA1851 Motherboard by CPU Support, M.2, USB and Overclocking
• CAMM2 and LPCAMM2: the third kind of laptop memory, and what it does and does not tell you about upgrades
• Check PC-Part Compatibility Before Buying: CPU, BIOS, RAM, Case and Power
• Integrated vs. Discrete Graphics: Check Workloads, CPU Models and Outputs
• Do you need a Copilot+ PC? The 40 TOPS / 16 GB / 256 GB requirement set, and whether your software runs on an Arm model

Comparisons: browse all guides →