AMD ships the firmware, your board maker ships the BIOS, and the version number on the download page belongs to neither one of them on its own
AGESA is AMD's firmware for the processor. AMD writes it, hands it to the board makers, and each board maker folds it into its own BIOS release for each of its own boards. XDA Developers puts the equivalence in one sentence: "Intel calls microcode by its name, while AMD calls it AGESA, but they're essentially the same thing."
So there are two version numbers in play, from two different companies, and they are not the same number. Learning to tell them apart is the whole skill, and it keeps working for the life of the platform. What this page will not do is tell you which AGESA version to install — here is why that is a feature, not a gap.
Your processor is not a fixed object. It needs code of its own to start, and that code can be replaced after the chip has been sold. XDA describes what that layer does: "Microcode is like a BIOS for your CPU." And what it is for: "It wakes the cores, sets the voltages, and gives AMD and Intel a way to issue updates to their hardware without recalling sold chips."
Both vendors have this. Only the names differ, and XDA states the mapping directly: "Intel calls microcode by its name, while AMD calls it AGESA, but they're essentially the same thing." If you have come from the Intel side and know what a microcode revision is, you already know what AGESA is; it is the same job under an AMD name.
HotHardware describes the AMD arrangement in more detail. First, that it is not optional: "Like all modern processors, AMD's Ryzen chips require microcode updates at boot to do their thing." Second, where it comes from and what it carries: "These microcode updates come from AMD as part of its AGESA firmware that contains System Management Unit (SMU) configurations for each type of processor supported by that particular mainboard."
Read that second sentence slowly, because two things in it decide everything below. AGESA comes from AMD, not from ASUS or MSI or GIGABYTE or ASRock. And what it contains is per-processor configuration, scoped to the processors that particular mainboard supports — which means the package your board gets is not identical to the package a different board gets.
You will see the acronym expanded in forum posts. We are not printing an expansion here, because neither of the two sources this page cites gives one, and AMD's own support pages did not respond when we tried to read them for this article. It is a name; nothing about reading a version string depends on knowing what it abbreviates.
The chain has three links, and the middle one is where readers lose track of which number is which.
XDA sets out what tends to be in one of those releases: "You'll normally see three things in a BIOS, in order from most to least common: improved memory compatibility, microcode or AGESA, and support for new CPUs." Note that microcode or AGESA is one of the three, not all of them — a BIOS release is a bundle, and its version number is a label for the bundle.
The easiest way to see that the two numbering schemes have nothing to do with each other is to look at a real vendor list. ASUS publishes the BIOS history for the ROG Strix B650E-E Gaming WiFi, a board we cover, on its support pages. Read on 15 September 2026, that list ran from version 1410, dated 2023/04/14, to version 3886, dated 2026/07/02.
Four digits, no dots, no platform prefix, no letter on the end. Nothing in that column resembles an AGESA string, because nothing in that column is one — those are ASUS's numbers for ASUS's builds. "My board is on 3886" and "my board is on AGESA something" are two different statements about two different things, and a forum thread that treats them as interchangeable is the reason people flash the wrong file.
When a vendor does name the AGESA inside a release, it looks like the examples HotHardware reported in December 2023. These are historical illustrations of the format, quoted with their date so nobody mistakes them for current advice: "AGESA version ComboAM5PI 1.0.8.0, released in early October, was the first to add support for Phoenix to the Socket AM5 platform," and, later in the same article, "The latest version, AGESA 1.1.0.1a, was released just last week."
Taking ComboAM5PI 1.0.8.0 apart:
ComboAM5PI. It names the socket family the package targets. HotHardware's example is explicitly about "the Socket AM5 platform", and the AM4 equivalent carries an AM4 prefix in the same position. This is the part to check first: a string with the wrong socket in it is not about your board at all.1.0.8.0. Four fields separated by dots, counting up over time. It is AMD's number, and it is the only part most people quote.a in 1.1.0.1a. A revision of an already-published number rather than a new one. A trailing letter is not a typo and not a different product line; it is the same release, amended.There is a fourth number that sits inside AGESA rather than in its name, and it explains why the outer number moving does not always tell you much. HotHardware: "the version number of the Phoenix SMU microcode has incremented rapidly, changing with every successive AGESA release." In that article the SMU code had reached "version 76.78.0". You will not normally need it, but if a release note quotes a number that looks nothing like either of the two above, this is the third kind.
This is the part no forum thread says, so it is worth being blunt about.
Nowhere on this site will you find a sentence naming the AGESA version your board should be running. Not on a review page, not on an upgrade page, not here. That is a standing editorial rule, it is enforced automatically across the corpus, and the reason is simple: published AGESA strings conflict with each other, and none of them is verifiable per board from where we sit. We can read what a vendor lists for one model on one day. We cannot read it for every model on every day, and a firmware recommendation that has quietly gone stale is worse than no recommendation, because it reads as though someone checked.
What follows from that is more useful than the number would have been:
People compare two boards, find they list the same AGESA, and expect identical behaviour. They do not get it, and the sources above explain why without any speculation being needed.
AGESA is one component of a release, not the release. XDA's list has memory compatibility as a separate line item from microcode or AGESA, and the most common of the three at that — so a vendor can ship a BIOS whose headline change is entirely its own work, with the AMD component unchanged. Two vendors doing that at different times, on top of the same AGESA, produce two boards that measurably differ.
And the AMD component itself is not one fixed thing either. HotHardware's wording is that AGESA carries SMU configurations "for each type of processor supported by that particular mainboard". The scope is per mainboard. A board with a different CPU support list is receiving a differently-scoped package under the same name.
On top of that sits everything the vendor decides: default settings, fan behaviour, how its firmware presents memory profiles, which of its own bugs it has fixed this month. None of that arrives from AMD. If you are trying to work out why two boards on the same chipset diverge, our chipset guide covers the part of the difference that is silicon rather than firmware.
There are real reasons, and they are the same reasons as for any BIOS update, because a newer AGESA only ever reaches you as one.
A processor the board does not yet know. This is the common one, and it is the third item on XDA's list of what a BIOS release contains: "support for new CPUs". A board built before a chip existed does not know how to start it. If you are asking whether your AM5 board will take a processor that has not shipped yet, that is an AGESA question at heart, and our Zen 6 on AM5 compatibility page is where we track what AMD has actually said — as opposed to what it is assumed will happen.
A fix that names your problem. Firmware is how a CPU-level defect gets corrected in the field. XDA's example is blunt: "Last generation, a slew of AMD Ryzen 7 7800X3D CPUs were burning themselves up in the CPU socket, again due to improper voltages. AMD addressed the issue with a new AGESA in a matter of days." That is the mechanism working as intended, and it is why "never update" is as wrong as "always update".
A security fix. These arrive the same way. HotHardware, on the December 2023 release: "the primary purpose of BIOS updates bearing this code was to address the serious LogoFAIL exploits". Nothing else on your machine will install that for you — not Windows Update, not a driver package.
Occasionally, performance. XDA: "Intel released microcode to improve the performance of Arrow Lake CPUs, as mentioned above, and AMD released a new AGESA to improve the lackluster performance of Ryzen 9000 CPUs when they were first released." Worth knowing that this happens, particularly on a platform in its first months. Not worth flashing speculatively in the hope that it will.
What is not on that list is a version number being larger than yours. If you have worked through the release notes between your version and the current one and none of them names something you are experiencing, you do not have a reason yet. The full procedure, including the parts that can leave you with a board that will not start, is in our BIOS update guide.
Three steps, in this order, and the first one is the one people skip.
We have not surveyed how each of the four board vendors writes its release notes, so this page does not promise that yours names an AGESA version at all. Some releases say little. Read what is actually in front of you for your model, and treat it as the answer — it outranks anything written about boards in general, including this page.
Every quoted sentence on this page comes from one of the three documents linked below, read on 15 September 2026. No board was flashed, opened or tested for this article, and nothing here is a benchmark or a measurement.
Three deliberate omissions. We name no AGESA version as the one your board should run, for the reason set out above. The two 2023 version strings quoted in Reading the Version String are HotHardware's, dated and attributed, and are there to show the format — they are not a recommendation and were already superseded long before this page was written. And we make no prediction about AGESA versions AMD has not released or CPU support AMD has not announced.
One source we could not use: AMD's own release-note pages did not respond on either the day this page was commissioned or the day it was written, so nothing here is sourced to AMD directly. Both articles cited below are reporting on AMD, which is a weaker thing, and we would rather say so than imply otherwise.
Published 15 September 2026. All three documents above were read on 15 September 2026.