Steam Frame OLED Headset: What Valve's Code Really Shows

Valve's own kernel source describes a display with 3,552 by 3,840 pixels per eye, built around a panel codenamed Plano. The only mode its code actually switches on, though, sends that panel just a quarter of those pixels, according to VR.org (yesterday). That mismatch is driving fresh speculation about a Steam Frame OLED headset, and the code itself tells a messier story than most headlines suggest.

Steam Frame Hub first spotted the Plano branch two days ago in deckard-linux, the public GitHub mirror of the Linux kernel that runs SteamOS on the headset (VR.org, yesterday). The Verge reported the find that same day, noting the panel's resolution and its Snapdragon 8 Gen 3 chip, the same processor already inside the shipping Frame, but did not tie either detail to OLED (The Verge, two days ago). VR.org's closer read of the code, published the following day, adds the detail that matters most: Plano's own driver never uses the word OLED anywhere (VR.org, yesterday).

Why untangle this now: the current Steam Frame only shipped last month, priced from $1,059 to $1,299 (Tom's Hardware, last month), and some reservation holders have already been told their order confirmation might not land until December, or into 2027 (VR.org, yesterday). For anyone weighing the Frame that exists today against one that might not, what's actually in Valve's kernel tree matters more than the rumor stacked on top of it.

Two clues, not one confirmed spec

The driver file panel-deckard-plano.c identifies itself as the "Plano DSI Panel Driver" and hardcodes resolution constants of 3,552 by 3,840, according to VR.org's review of the source (VR.org, yesterday). Its device-tree files target the SM8650, Qualcomm's part number for the Snapdragon 8 Gen 3 already running inside the shipping Frame, and its testing notes reference development motherboards Valve calls EV2 and DV2, the same codenames used for the current headset's own board revisions (VR.org, yesterday).

None of that confirms the Steam Frame Plano panel runs on OLED. A separate commit, dated December 17, 2025, and titled "create new deckard-panel for oled display," does use the word, but it lives in its own file and runs at 1,776 by 1,920, far below Plano's resolution. Plano itself was added to the codebase on March 5, 2026, as a distinct driver with no reference back to the OLED commit (VR.org, yesterday).

Put plainly: the highest-resolution driver doesn't claim OLED, and the driver that claims OLED doesn't claim high resolution. "Steam Frame OLED headset" is currently an inference built by stacking two unrelated commits together, not a specification Valve has written down anywhere. Valve has not commented on either file (VR.org, yesterday). Road to VR has noted that Deckard, the Frame's own internal codename, surfaced in leaked code the same way back in 2021, years before the headset actually shipped, a useful reminder that codenames in a kernel tree aren't launch announcements (VR.org, yesterday).

The rumored Steam Frame OLED resolution isn't arbitrary, though. VR.org notes that 3,552 by 3,840 per eye matches the micro-OLED panels in Samsung's $1,799 Galaxy XR and the Sony panels Pimax specifies for its Dream Air, which suggests Valve may be testing against glass that already exists in the supply chain rather than designing something bespoke (VR.org, yesterday). Valve watcher Brad Lynch, quoted by The Verge, floated a similar read: Valve could be "just trying to prototype headsets with this optical stack based on what's already lying around" rather than finalizing a product (The Verge, two days ago).

Steam Frame OLED vs LCD: why only one transport mode works

Strip away the speculation and the working code describes something narrower than a next-gen display. The one enabled mode accepts 1,776 by 1,920 pixels per eye at 72, 80 or 90Hz, then relies on a panel-side chip Valve calls Global 2x to double that image in each direction and fill the full display (VR.org, yesterday).

That working input totals roughly 3.4 million pixels per eye, about 27% less image data than the 4.7 million pixels per eye the shipping LCD Frame already handles (VR.org, yesterday). It's a real regression, not just a smaller number on paper: the shipping Frame already runs up to 120Hz, with 144Hz labeled experimental (Tom's Hardware, last month), while Plano's only working mode tops out at 90Hz, and its own 100, 110 and 120Hz entries are marked "TEST ONLY" and "not working" (VR.org, yesterday).

Other documented modes go further but aren't finished. A 1.5x mode, at 2,368 by 2,560 per eye, is logged as working with low-level display tools "but not steamvr," and separately noted to display only the left eye while the system bogs down (VR.org, yesterday). Full native resolution on both eyes at 90Hz, which would move roughly 2.5 billion pixels per second, about three times the shipping Frame's estimated 840 million, is marked "working for dphy" but carries its own caveat: "requires MIPI overclocking" (VR.org, yesterday).

Those specific examples point to an unfinished display-link implementation rather than a finished product sitting on a shelf. Whether the panel itself, the transport, or some combination ends up limiting a future headset isn't something the code resolves on its own. What is consistent across the entries VR.org documented is that the simplest option is the only one that works, and the denser ones remain labeled broken, untested, or dependent on pushing the hardware past its stated limits.

Several unfinished paths, no chosen solution

The code shows Valve testing more than one way to feed a denser panel over a limited connection, and none of them reads as settled. Display Stream Compression, a standard method for squeezing more image data through a limited link, shows up across several mode entries (VR.org, yesterday). The only mode that actually works, Global 2x scaling, is the simplest of the group: send fewer pixels and let a chip on the panel enlarge them (VR.org, yesterday).

Foveal modes, which would send full detail only to wherever the eye is pointed on the display itself, appear least complete of all. VR.org found that the "plumbing" needed to deliver that kind of targeted patch to the panel simply isn't built yet (VR.org, yesterday). That's worth separating from foveated streaming, a feature the shipping Frame already uses over its wireless PC link, drawing on its internal eye-tracking cameras to cut network bandwidth rather than panel bandwidth (Tom's Hardware, last month; Ars Technica, 11 months ago). The code shows Valve experimenting with scaling, compression and foveation as possible paths to a denser display, but it treats none of them as finished or chosen.

What this means for Steam Frame buyers

For anyone shopping today, the calculus doesn't shift. The current Frame is a real, reviewed product: 2,160 by 2,160 LCD panels per eye, a Snapdragon 8 Gen 3 chip, 16GB of RAM, and a 440g weight built around a modular 185g core, well under the 809g of the original Valve Index (Tom's Hardware, last month; Ars Technica, 11 months ago). Tom's Hardware called wireless streaming performance "excellent," while noting standalone performance is held back mainly by the overhead of translating x86 Windows games to run on the headset's Arm chip (Tom's Hardware, last month).

None of that changes based on Plano. What the code doesn't establish is a price, a launch window, or even a confirmed OLED product. The only public timeline is a guess from Steam Frame Hub of "late 2027 at the earliest," explicitly labeled speculation rather than anything resembling a Valve roadmap (VR.org, yesterday). For a sense of Valve's pacing on hardware refreshes, the company took more than a year and a half to follow the original Steam Deck with the Steam Deck OLED (The Verge, two days ago).

A more concrete near-term change is coming on the accessory side. Valve has confirmed, through Road to VR citing Skarred Ghost, that it is developing a new headstrap with hot-swappable batteries and BMR audio drivers to match the Index's speakers, though it has given no price, battery capacity, or ship date (VR.org, yesterday). That's a confirmed, in-development upgrade, distinct from Plano's unconfirmed display status, and it bears on a real limitation: battery life currently runs about an hour in native Half-Life: Alyx and more than double that when streaming from a PC (Tom's Hardware, last month).

Reservation holders are working on a separate timeline entirely. Valve told some of them last week that an order confirmation email may not arrive until December or sometime in 2027 (VR.org, yesterday). For that group, whether to hold out for OLED is largely beside the point: a wait is already built into the order, regardless of what Plano eventually turns out to be.

What happens next

The most telling detail in Valve's code isn't the resolution number at all. It's that the simplest transport trick, 2x upscaling, is the only one that currently works, while the specific higher-fidelity modes VR.org documented sit behind a "test only," a "not working," or a requirement to overclock the display link (VR.org, yesterday).

Valve has confirmed exactly one near-term hardware change, the swappable-battery strap, and said nothing publicly about Plano, OLED, or a second-generation Frame (VR.org, yesterday). What the kernel tree shows is an unfinished display-link implementation that Valve is actively testing on hardware it already ships, not proof of a product waiting in the wings. Anyone deciding whether to buy is better off judging the Frame that exists today than the one hinted at in a commit log.