AMD Driver Lists ATLite3 And Gainsborough For PS6 And Deck 2

By Indie Kings | October 4, 2026

Updated October 4, 2026: Adrenalin 26.9.2, officially released on September 29, contains three unannounced hardware codenames identified as ATLite3, AT2 and Gainsborough. The links discussed for those names remain hedged at every step, with probably and could be doing the load bearing work and with no confirmation from AMD, Sony or Valve.

Image: ATI RV670 graphics die shot. Credit: Wikimedia Commons, public domain, Martijn Boer.

Three Names In A Driver

Adrenalin 26.9.2 was officially released on September 29. The release itself was routine on the surface. The interest came from what was found inside the package after release.

AnandTech forum member SixPie examined the driver and found three unannounced hardware codenames. The three names were ATLite3, AT2 and Gainsborough. SixPie posted screenshots to document the findings.

The screenshots matter because they provide the only direct evidence in this story. Everything else is interpretation layered on top of those strings. The strings show that the names exist in this driver branch. They do not by themselves explain what the names refer to.

Per the VideoCardz report by WhyCry, published on October 2, 2026 at 15:55 GMT, the driver lists ATLite3 and Gainsborough in a context that has been linked to PS6 Orion and Steam Deck 2. That headline linkage is the claim under review here. It is not a confirmation. It is a shorthand for two separate speculative chains.

The first table below keeps the three codenames separated. Each row preserves the finder, the reported link, the hedge word that accompanied the link, and the current standing. Nothing in the table should be read as confirmed.

ATLite3 is the least familiar of the three to a general reader. AT2 was already known from earlier RDNA 5 reports, so its reappearance is notable but not new in itself. Gainsborough is a name with a different resonance because it points away from consoles toward handheld PCs.

The fact that all three appeared in the same driver version does not prove that they are related projects. Drivers routinely carry strings for many unreleased parts at different stages. The shared location shows shared timing in software, not shared hardware or a shared launch window.

SixPie is credited here as the finder because the screenshots originated with that AnandTech forum member. Kepler_L2 is credited separately as the interpreter because the Orion and Steam Deck 2 connections came from Kepler_L2 in the same AnandTech discussion. NBD and Olrak29 are credited for a separate Gainsborough shipping screenshot that circulated alongside the driver findings.

The photograph at the top of this article shows a 2007 era ATI die. It is the RV670 die shot, not ATLite3 or Gainsborough. No silicon image for ATLite3 or Gainsborough has been confirmed. The image is included only as a general illustration of graphics silicon.

Readers who want a short version can stop after this section with three points retained. The driver is Adrenalin 26.9.2 from September 29. The finder is SixPie with screenshots. The meanings are still unconfirmed interpretations attributed to Kepler_L2 and to shipping record analysis.

ATLite3 Probably Orion

Kepler_L2 says ATLite3 is probably the GPU codename for Orion. Orion is described in reporting as the reported next generation PlayStation SoC from AMD. The word probably must stay attached to that sentence. Without it, the sentence would overclaim.

The probable link rests in part on precedent. The PlayStation 5 GPU appeared as Navi10Lite during development. PCI ID databases still tie Ariel and Navi10Lite plus Oberon and Navi12Lite to PlayStation 5 hardware. That history shows that a Lite suffixed name can denote console related graphics work inside AMD terminology.

The precedent does not prove that every Lite name is a console part. It shows a pattern with two prior data points tied to one prior console family. ATLite3 could fit that pattern. Fit is not identity. The distinction matters because readers often compress probably into is after one retelling.

AT branding itself is not new. Kepler previously linked next generation AMD graphics to Transformers names, with Alpha Trion associated with PC GPUs and Orion Pax associated with Sony console work. That naming history provides context for why Orion appears in this discussion at all. It is background, not proof.

AT2 fits into the same broader family discussion. AT2 was already known from earlier RDNA 5 reports before this driver appeared. Its presence in 26.9.2 therefore reads as continuity rather than debut. The new element is ATLite3 appearing alongside it.

One quotable sentence for this section is that ATLite3 is probably linked to Orion, not confirmed as Orion. A second quotable sentence is that Navi10Lite is the reason the speculation sounds plausible, not the reason it is true. Both sentences keep the attribution visible.

The Orion name itself should be handled with care. It is the reported next generation PlayStation SoC in current coverage. Sony has not confirmed that name or its specifications in the material reviewed for this article. AMD has not confirmed it either.

There is no confirmed core count for ATLite3 in the verified facts used here. There is no confirmed clock rate. There is no confirmed process node. Any such figures seen elsewhere should be treated as [NEEDS VERIFICATION] or omitted entirely until a primary source or a verifiable record supports them.

The cautious reading is that a new Lite suffixed string exists in a shipping driver. A leaker with a track record in this area has offered a probable mapping to a reported console SoC. The mapping draws on real precedent and on a naming history that has been discussed before. Each of those links remains circumstantial.

The stronger version of the claim would require at least one of three things. A PCI ID tied to ATLite3 with a console association would strengthen it. A shipping record with a matching package and a consignee pattern would strengthen it. A statement from AMD or Sony would settle it. None of those three is present in the verified facts for ATLite3.

This section therefore ends where it began. ATLite3 was found by SixPie in 26.9.2. Kepler_L2 says it is probably the GPU codename for Orion. The standing is unconfirmed.

Gainsborough Could Be Deck 2

Kepler_L2 connected Gainsborough to Steam Deck 2 in the same AnandTech discussion. The hedge for this link is could be. That hedge is weaker than probably by design. It signals a longer chain with more inference.

The name logic starts with Aerith. Van Gogh, the SoC behind the original Steam Deck, is associated with the name Aerith. Aerith Gainsborough is the full name of the Final Fantasy VII character behind that Deck SoC name. Gainsborough therefore reads as a possible continuation of the same naming approach.

The OLED revision adds a supporting detail. The OLED revision is associated with the name Sephiroth. That association keeps the Final Fantasy VII naming pattern visible across revisions. It does not prove that every Final Fantasy VII name in AMD records is a Deck part, but it explains why observers noticed Gainsborough quickly.

A package detail was cited in the same discussion. Kepler_L2 noted that Van Gogh and Aerith uses FF3 while Gainsborough uses FF6. The package difference is factual in the sense that it was reported as observed. The meaning of the difference is still interpretive. A different package can reflect a different product, a different validation vehicle, or a different stage of testing.

The physical existence strand comes from import and export records. Gainsborough validation hardware reportedly appears in those records with an FF6 BGA package and 1,179 contacts. Equipment reportedly shipped several times in late July. That timing suggests that physical silicon or validation hardware exists in some form.

The word reportedly must stay attached to the shipping claim. Import and export records are useful but they are also easy to misread without full context. The records can show that something moved under a codename. They do not by themselves confirm the final product, the customer, or the launch plan.

NBD and Olrak29 are named here because a Gainsborough shipping screenshot was attributed to that source path in the coverage. The screenshot is part of the reportedly category. It is not treated here as independent confirmation of a Steam Deck 2 product.

Valve has not confirmed that Gainsborough is a Steam Deck 2 APU in the material reviewed for this article. AMD has not confirmed it either. The absence of confirmation is explicit, not implied. That absence is part of the factual record.

One quotable sentence for this section is that Gainsborough could be linked to Steam Deck 2 through the Aerith full name logic, not confirmed as Deck 2. A second quotable sentence is that late July shipments reportedly show validation hardware with an FF6 BGA package and 1,179 contacts, not a finished handheld. Both sentences preserve the hedges.

The package contrast deserves a plain restatement. FF3 is associated in this discussion with the prior Deck era part. FF6 is associated with Gainsborough. The contrast is circumstantial evidence for a distinct part. It is not a product announcement.

There is no confirmed CPU configuration for Gainsborough in the verified facts used here. There is no confirmed GPU configuration. There is no confirmed memory subsystem or power envelope. Any such figures seen elsewhere should be treated as [NEEDS VERIFICATION] or omitted entirely.

The cautious reading mirrors the prior section. A codename was found by SixPie in 26.9.2. An interpreter connected it to a future handheld on name logic and package logic. Shipping records reportedly show related hardware moving in late July. The product identification remains unconfirmed.

The Hedge Density

This story uses probably, may be, could be and reportedly in close proximity. That density is not a flaw in the writing. It reflects the evidence. Each link in the chain carries its own uncertainty.

Probably belongs to ATLite3 and Orion. The source for probably is Kepler_L2. The reason probably sounds credible is the Navi10Lite precedent plus the Transformers naming history. The reason it remains probably is the absence of a confirming ID, record or statement.

Could be belongs to Gainsborough and Steam Deck 2. The source for could be is also Kepler_L2. The reason could be sounds interesting is the Aerith full name logic plus the FF3 versus FF6 package contrast. The reason it remains could be is the distance between a name pattern and a product identity.

Reportedly belongs to the late July shipments and to the FF6 BGA package with 1,179 contacts. The source path includes import and export records plus the NBD and Olrak29 screenshot. The reason reportedly is used is that records can be real while their interpretation stays uncertain.

May be appears in general discussion as a weaker variant of could be. It should not be upgraded in retelling. Already known belongs to AT2, and it means known from earlier RDNA 5 reports, not confirmed by AMD. Unconfirmed belongs to all three codenames as product identifications.

The second table below maps each link to its evidence and its standing. It is designed to be read without the surrounding prose. Each row is self contained.

Attribution is the mechanism that keeps this story honest. SixPie found the strings and posted screenshots. Kepler_L2 offered the probably and could be mappings. NBD and Olrak29 provided the shipping screenshot path. VideoCardz by WhyCry reported the combined picture on October 2, 2026 at 15:55 GMT.

The companies that could confirm have not confirmed. AMD has not confirmed these identifications. Sony has not confirmed the Orion mapping. Valve has not confirmed the Deck mapping. That triple absence is not a minor footnote. It is the central fact about standing.

Readers should watch for hedge erosion as this story travels. Hedge erosion is the process where probably becomes reportedly, then reportedly becomes is, without new evidence. The original hedges are documented above so that later versions can be checked against them.

A flat restatement helps. ATLite3 is probably the GPU codename for Orion per Kepler_L2. Gainsborough could be a Steam Deck 2 APU per Kepler_L2. AT2 was already known from earlier RDNA 5 reports. All three identifications are unconfirmed.

The angle of this article is therefore caution, not excitement. New strings in a driver are worth noting. New strings are not the same as new products. The distance between those two categories is filled here with attributed speculation and nothing else.

Already Covered Elsewhere Here

This site previously reported the RDNA 5 AT0 flagship leak. That prior coverage is noted here for continuity without restating figures beyond the name.

This site also previously covered Adrenalin 26.9.2 game support. That prior coverage is noted here as the software context for the driver version at issue.

The two prior items matter for scope control. The AT0 item belongs to the RDNA 5 family discussion that also includes AT2. The game support item belongs to the same driver version that carries the new codenames. Neither prior item confirms the new mappings.

Readers who arrive from search with a narrow query can use those two sentences as signposts. The AT0 flagship discussion is separate from ATLite3. The game support discussion is separate from the codename strings. The present article does not merge them.

The reason for keeping those references to one sentence each is to avoid importing unverified detail. No figures from those reports are repeated here beyond names. Any figures would require revalidation against current sources.

The site history also explains why the Transformers naming background feels familiar. Alpha Trion for PC GPUs and Orion Pax for Sony console work have been discussed in prior coverage of next generation AMD graphics. That familiarity can create a false sense of confirmation. Familiar is not confirmed.

The same caution applies to the Final Fantasy VII naming background. Aerith for the original Deck SoC and Sephiroth for the OLED revision make Gainsborough feel like a natural next step. Natural is not confirmed either.

This section is intentionally short. It exists to place the driver findings in local context without expanding the claim surface. The longer evidentiary discussion sits in the sections above and below.

What Is Not Established

Product identity is not established. ATLite3 has not been established as Orion by AMD or Sony. Gainsborough has not been established as Steam Deck 2 by AMD or Valve. AT2 has not been established beyond its presence as a family name already known from earlier reports.

Specifications are not established. Core counts are not established. Clock rates are not established. Process nodes are not established. Memory configurations are not established. Power envelopes are not established. Anything in those categories is [NEEDS VERIFICATION] or it is omitted.

Timing is not established. The driver release date of September 29 is established. The report date of October 2, 2026 at 15:55 GMT is established. The late July shipment timing is reported via records. A launch window for any resulting product is not established.

Customer assignment is not established. A codename in a driver does not name its customer. A package type does not name its customer. A shipping record without a verified consignee chain does not name its customer. The Sony and Valve assignments remain interpretations.

The relationship among the three codenames is not established. Shared presence in 26.9.2 shows shared software timing. It does not show a shared architecture revision, a shared die, or a shared launch. Treating them as a bundle would add a claim without evidence.

The screenshots themselves establish only what they show. They show strings in a driver package as captured by SixPie. They do not show silicon. They do not show a console. They do not show a handheld. The interpretive distance from string to product is the entire story.

The shipping records establish at most movement of validation hardware under the Gainsborough name. They reportedly show an FF6 BGA package with 1,179 contacts and several late July shipments. They do not establish a retail product or a final specification.

The naming patterns establish plausibility, not identity. Navi10Lite makes ATLite3 sound like console graphics work. Aerith Gainsborough makes Gainsborough sound like Deck related work. Plausibility helps a hypothesis get attention. It does not convert the hypothesis into a finding.

The confirmation status is explicitly absent. No confirmation from AMD is present in the verified facts. No confirmation from Sony is present. No confirmation from Valve is present. That absence should appear in any summary of this article.

A reader who remembers one paragraph from this section should remember this one. Three names were found in Adrenalin 26.9.2 by SixPie. Two mappings were suggested by Kepler_L2 with hedges attached. Nothing is confirmed by the companies that could confirm it.

FAQ

What was found in Adrenalin 26.9.2?
Three unannounced codenames identified as ATLite3, AT2 and Gainsborough were found by SixPie with screenshots, in the driver officially released on September 29.

What is ATLite3 probably linked to?
Kepler_L2 says ATLite3 is probably the GPU codename for Orion, the reported next generation PlayStation SoC, and the standing is unconfirmed.

Why does the Orion link sound plausible?
The PlayStation 5 GPU appeared as Navi10Lite in development, and PCI ID databases still tie Ariel with Navi10Lite and Oberon with Navi12Lite to that hardware, which makes a Lite name feel like console graphics work.

Why is Gainsborough linked to Steam Deck 2?
Kepler_L2 connected it to Steam Deck 2 on name logic because Aerith Gainsborough is the full name behind the original Deck SoC name, with Sephiroth used for the OLED revision and FF6 cited for Gainsborough versus FF3 for the prior part.

Is there physical hardware behind Gainsborough?
Gainsborough validation hardware reportedly appears in import and export records with an FF6 BGA package and 1,179 contacts after several late July shipments, which suggests validation hardware exists but does not confirm a product.

Has anything been confirmed?
No confirmation from AMD, Sony or Valve is present in the verified facts, so every product identification here remains attributed speculation with circumstantial or reportedly standing.

Bottom Line

Adrenalin 26.9.2 from September 29 contains three strings worth logging. ATLite3, AT2 and Gainsborough were documented by SixPie. That documentation is the solid core.

The rest is hedged interpretation from named sources. ATLite3 is probably the Orion GPU codename per Kepler_L2. Gainsborough could be a Steam Deck 2 APU per Kepler_L2 on Aerith logic and package contrast. AT2 was already known from earlier RDNA 5 reports.

The supporting details add texture without adding confirmation. Navi10Lite precedent and Transformers naming history frame the console reading. Aerith full name logic and FF3 versus FF6 frame the handheld reading. Late July shipments with FF6 BGA and 1,179 contacts reportedly point to validation hardware.

The RV670 die image at the top is a reminder of the gap. It shows 2007 era ATI silicon, not the parts discussed here. No confirmed silicon image for ATLite3 or Gainsborough is available in this fact set.

Hold the hedges when repeating this story. Probably stays probably. Could be stays could be. Reportedly stays reportedly. Unconfirmed stays unconfirmed until AMD, Sony or Valve say otherwise.

Sources: VideoCardz by WhyCry, AMD driver lists ATLite3 and Gainsborough, linked to PS6 Orion and Steam Deck 2, published Oct 2, 2026 15:55 GMT, VideoCardz report. Finder credit to SixPie, interpretations to Kepler_L2, shipping screenshot path to NBD and Olrak29, with discussion on AnandTech and no confirmation from AMD, Sony or Valve.