Epic Games says the next version of its PC storefront software is not just a visual refresh. On October 14, 2026, the company pushed a rebuilt Epic Games Launcher into Preview and claimed it opens to the store page 46% faster than Steam from a cold start, a figure that landed across gaming outlets within a day and reignited the decade-long rivalry between the two largest PC game storefronts. The twist that made the story travel further than a typical launcher update: Epic says the new client was built using AI-assisted “vibe coding” rather than a conventional line-by-line rewrite.
GamesIndustry.biz reported that Epic’s benchmark, run by contracted testing firm Keywords Studios across 13 separate machines, found the new launcher reached the store page in an average of 5.47 seconds, against 10.07 seconds for Steam under the same conditions. Epic also claims a 14% edge restoring the client from the system tray to the library view: 0.36 seconds versus 0.42 seconds for Steam. Those numbers, thin as the second comparison is in absolute terms, are now the center of a fresh argument about whether AI-generated software can out-engineer a storefront that has spent more than two decades optimizing for exactly this kind of friction.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Epic Games Actually Shipped on October 14, 2026
The release itself is modest in scope even if the marketing around it is loud. Epic Games launcher vs Steam speed claims aside, what actually went live is a Preview build, meaning it runs alongside the existing launcher rather than replacing it outright. The Preview is available for Windows 10 and 11 and macOS 12 or newer, and for the first time Epic is shipping a native Linux client, a platform the company had never directly supported with its own launcher before. The Verge covered the Linux expansion as one of the more consequential parts of the announcement, since it opens Epic’s storefront to SteamOS and other Linux-based gaming setups without relying on compatibility layers.
Epic’s own framing of the rebuild leaned into the AI angle rather than downplaying it. In the announcement, the company stated: “Programming is hard. So, instead of doing that, we vibe coded a new Epic Games Launcher that’s way faster and more responsive than the current one, and has a new look.” That is an unusually blunt admission from a major platform holder, and it is the line that most outlets, including Tom’s Hardware, used as the hook for their coverage.
Epic also described how the code was actually produced, saying AI agents wrote and reviewed much of the code while engineers supervised the process. That detail matters for anyone trying to gauge how much human oversight sat between the AI output and the shipped binary, since it positions the rebuild as supervised AI-assisted development rather than a fully autonomous process.
The Benchmark Numbers Behind the 46% Claim
The specific figures matter because they are the entire basis for the “46% faster than Steam” headline that spread across tech and gaming press on October 15. Epic’s test methodology, as described in the announcement and cited by multiple outlets, averaged five test runs on each of 13 machines, excluding the three lowest-end machines from the final figures. That exclusion is worth flagging on its own: dropping the weakest hardware from a performance comparison tends to flatter whichever client benefits most from a stronger baseline, and neither Epic nor Keywords Studios has published the full, unfiltered dataset.
| Benchmark | Epic Games Launcher (new) | Steam | Epic’s claimed advantage |
|---|---|---|---|
| Cold start to store page | 5.47 seconds | 10.07 seconds | 46% faster |
| System-tray restore to library | 0.36 seconds | 0.42 seconds | 14% faster |
| Library load time vs. previous Epic launcher | Reported as 30x faster | Not applicable | Internal comparison only |
| Startup time vs. previous Epic launcher | Reported as roughly 2x faster | Not applicable | Internal comparison only |
| Test sample | 5 runs each on 13 machines; 3 lowest-end machines excluded, per Epic Games | ||
Tom’s Hardware pointed out a detail that a lot of the breathless coverage glossed over: the 14% tray-restoration advantage amounts to 0.06 seconds in absolute terms, a gap unlikely to register for the overwhelming majority of players. The outlet’s own headline framed the Preview as not recommended for competitive players, a caution that Epic itself has echoed by labeling the build “Preview” rather than stable. In practice, that means the comparison that is driving headlines today is a controlled lab benchmark on specific hardware, not a claim that has been independently reproduced by a third party outside Epic’s own contracted testing firm.
Martin Keely on the Rebuild: Library Loads, Startup Speed
Martin Keely, vice president and general manager of the Epic Games Store, has become the public face of the rebuild in interviews following the Preview launch. Speaking about the launcher’s internal performance gains, Keely told IGN: “The library loads 30 times faster.” That figure compares the new client against Epic’s own previous launcher rather than against Steam, a distinction that is easy to miss when it gets quoted alongside the Steam-facing 46% number but is methodologically a separate claim entirely.
Keely also discussed startup performance in comments reported by GamesIndustry.biz, saying: “The startup is twice as fast.” Here again, the comparison is against the previous Epic Games Launcher, not Steam, which is a reasonable basis for an internal engineering update but a looser standard than the direct head-to-head cold-start figure that is doing most of the marketing work in this story.
Taken together, the Keely quotes paint a picture of a team that rewrote the launcher from near-zero rather than patching the existing codebase, which lines up with Epic’s own description of the AI-assisted rebuild. Whether a 30x internal improvement and a 46% external improvement over Steam can both be true at once depends heavily on how slow the old Epic launcher actually was relative to Steam’s baseline, something neither company has published a direct comparison for.
Why “Vibe Coding” a Storefront Is a Bigger Deal Than It Sounds
Vibe coding, the practice of describing a desired outcome in natural language and letting an AI model generate and iterate the resulting code with limited manual review, has mostly lived in indie-developer and hobbyist circles until now. Epic Games Launcher’s rebuild is the highest-profile example yet of a major platform holder applying that approach to commerce-critical infrastructure: software that handles purchases, account logins, and game library access for tens of millions of PC gamers.
That is a meaningfully different risk profile than vibe-coding a solo indie game’s gameplay scripts. A storefront client touches payment flows, DRM validation, and account security, which is why Epic’s own acknowledgment that “AI agents wrote and reviewed much of the code while engineers supervised the process” is doing real work in the announcement: it is a hedge against the obvious follow-up question about who is actually responsible for catching a security flaw if the AI-generated code introduces one. Industry commentary around vibe coding’s spread into game development has already flagged this tension between shipping speed and traditional code-review rigor, and Epic’s launcher is now the most concrete test case available.
Eurogamer’s framing of the release as Epic “throwing shade” at Steam captures the competitive subtext well: this is as much a marketing document aimed at developers and players who have complained about the old launcher’s sluggishness as it is a genuine engineering case study. Epic has spent years trying to close the gap with Steam on everything from library management to social features, and a headline-grabbing speed claim, true or not in every configuration, is a cheap way to reframe the conversation in Epic’s favor.
Historical Context: The Epic vs. Steam Storefront War
This launcher rebuild does not exist in a vacuum. Epic Games Store launched in December 2018 with a direct challenge to Steam’s long-standing 30% revenue cut, offering developers an 88/12 split and a rotating schedule of free games to build its user base quickly. That strategy bought Epic a foothold but never seriously dented Steam’s concurrent-user dominance; Steam has continued setting fresh concurrency records even as Epic’s free-game giveaways and exclusivity deals piled up.
Launcher performance has been one of the more consistent player complaints about Epic’s storefront since day one. Reviewers and forum threads have repeatedly flagged slow startup times, sluggish library loading, and a client that felt heavier than Steam’s for years, even as Epic added features like cloud saves and an achievements system to close functional gaps. The October 14 Preview release is explicitly aimed at that specific, long-running criticism rather than at Epic’s pricing model or exclusivity strategy, which suggests Epic sees raw client performance as the remaining lever it has not yet pulled hard on.
It is also worth noting how this fits into the current climate around PC gaming hardware and storefronts more broadly. With Valve pushing its own Steam Machine and Steam Frame hardware and expanding its device lineup beyond the original Steam Deck, Steam’s dominance is no longer just about the storefront software, it is increasingly about an entire hardware and software ecosystem that Epic has no direct equivalent for. A faster launcher narrows one specific gap; it does not touch the much larger one.
Market Impact: Does a Faster Launcher Move Market Share?
The honest answer, based on the coverage available so far, is that nobody has shown it does. The GamesIndustry.biz report and every outlet that followed it measured launcher startup and restoration speed, not sales, active users, playtime, or revenue. Epic has not published any data suggesting the Preview release has changed user acquisition, retention, or developer sign-up numbers, and it would be premature for any of that to show up this quickly after an October 14 Preview launch anyway.
What the claim does accomplish is a shift in the terms of the competition. For years, Epic vs. Steam coverage centered on revenue splits, exclusivity deals, and free-game giveaways. A benchmark-driven headline about raw client speed reframes Epic’s pitch around engineering velocity instead, which is a narrative Epic can control more directly than it can control developer loyalty to Steam’s established audience. Whether that reframing survives contact with ordinary players running the Preview on their own machines, outside Keywords Studios’ controlled 13-machine sample, is the open question analysts and players alike are now watching for.
Digital Trends’ coverage emphasized that Epic built the new client on an entirely new architecture rather than simply optimizing the old one, which is a more defensible technical claim than the headline speed numbers alone. If true, it suggests Epic is making a longer-term infrastructure bet rather than chasing a one-time marketing win, and the real test will be whether that new architecture holds up as Epic adds back the features the Preview build is still missing.
Competitive Comparison: Epic Games Launcher vs. Steam
| Factor | Epic Games Launcher (new Preview) | Steam |
|---|---|---|
| Supported OS | Windows 10/11, macOS 12+, Linux (new) | Windows, macOS, Linux (native, long-standing) |
| Cold start to store page (Epic’s benchmark) | 5.47 seconds | 10.07 seconds |
| Revenue split with developers | 88/12 (Epic’s established model) | Standard 70/30, tiered at higher revenue |
| Hardware ecosystem | None (software storefront only) | Steam Deck, Steam Machine, Steam Frame |
| Development approach for latest client | AI-assisted “vibe coding,” Epic says, with engineer supervision | Traditional development (no public AI-rebuild claim) |
| Release stage as of October 15, 2026 | Preview, not recommended for competitive use per Tom’s Hardware | Stable, long-established |
Industry Reaction and the Missing Independent Analysis
One notable gap in the coverage so far is the absence of independent analyst commentary on whether AI-generated client code of this kind is reliable or maintainable over the long run. The available reporting largely quotes Epic executives directly or describes the editorial framing of outlets covering the release, but no named security researcher, software engineering analyst, or competing platform executive has gone on record assessing the vibe-coded launcher’s underlying code quality. Valve, for its part, has not issued any public response to Epic’s benchmark or to the “vibe coded” framing, according to the coverage reviewed.
That silence from Valve is probably strategic rather than dismissive. Steam does not need to respond to every Epic marketing push, and a direct rebuttal would only amplify a story that is, for now, confined mostly to gaming and tech press rather than mainstream coverage. The more useful signal to watch for is whether other platform holders, including console makers or rival PC storefronts, start describing their own tools as AI-assisted in the way Epic has, since that would suggest the “vibe coded” framing is becoming a marketing category of its own rather than a one-off Epic talking point.
What’s Still Missing From the Preview Build
Preview releases are, by definition, incomplete, and Epic has been upfront that the new launcher is lighter on features than the current stable client while the company works toward a full release. Epic has asked users testing the Preview to report bugs as it continues to build out functionality, and the company has not published a public roadmap or date for when the rebuilt launcher will replace the existing stable version. No pricing or subscription changes have been reported alongside the release, and the Preview is explicitly an optional, parallel install rather than a forced upgrade.
That incompleteness is also why outlets like Tom’s Hardware cautioned against relying on the Preview for anything resembling competitive or mission-critical use. A storefront client that is still filling in missing features is not the same as a finished product, regardless of how favorable its benchmark numbers look in a controlled test.
Predictions: Where This Goes From Here
- Expect Epic to publish additional benchmark comparisons as the Preview build matures, likely expanding beyond cold-start and tray-restore metrics to library browsing, download speeds, and update installation times, the areas where Steam has historically had the most room for a direct comparison.
- Independent reviewers and YouTube hardware channels will almost certainly run their own side-by-side tests on a wider range of consumer hardware once the Preview is more widely installed, which will either validate or puncture the 46% headline figure outside Keywords Studios’ controlled sample.
- Linux support is likely to become a bigger talking point over the coming months as Valve’s own Linux-based hardware push continues; Epic having a native Linux client removes one of the long-standing reasons SteamOS and Linux gaming enthusiasts avoided the Epic Games Store entirely.
- Other platform holders and storefront operators will likely face pressure to disclose or deny their own use of AI-assisted development in client software, now that Epic has made “vibe coded” a visible marketing term rather than an internal engineering detail.
- The real test of this story’s staying power will be whether Epic’s stable release, whenever it ships, actually delivers on the Preview’s benchmark claims for ordinary users, or whether the final numbers settle closer to Steam’s existing performance once missing features are added back in.
What This Means for PC Gamers Right Now
For most players, the immediate, practical takeaway is narrow: the new Epic Games Launcher Preview is available to try today on Windows, macOS, and Linux, but it is not a replacement for the stable client yet, and nobody outside Epic’s own contracted test has independently verified the headline speed claims at this point. Players curious about the rebuild can install the Preview alongside their existing launcher without losing access to their library, which is the lowest-risk way to see whether the claimed speed gains hold up on their own hardware.
For developers and the broader gaming industry, the more interesting story is the precedent: a major platform holder publicly describing a core piece of commerce infrastructure as AI-assisted “vibe coding,” with specific, if narrow, performance claims attached. Whether that framing ages well depends entirely on how the stable release performs once it ships, and whether the cost and headcount pressures already reshaping parts of the games industry push more studios and platforms toward AI-assisted development regardless of how this particular launcher story plays out.
Frequently Asked Questions
What did Epic Games actually claim about its new launcher?
Epic says its rebuilt Epic Games Launcher, released in Preview on October 14, 2026, opens to the store page 46% faster than Steam from a cold start (5.47 seconds versus 10.07 seconds) and restores from the system tray 14% faster (0.36 seconds versus 0.42 seconds), based on testing by Keywords Studios across 13 machines.
What does “vibe coded” mean in this context?
Epic used the term to describe building the new launcher with AI-assisted development, where AI agents wrote and reviewed much of the code while Epic’s engineers supervised the process, rather than a traditional manual rewrite.
Is the new Epic Games Launcher available now?
Yes, as a Preview build for Windows 10 and 11, macOS 12 or newer, and for the first time, Linux. It runs separately from the existing stable launcher rather than replacing it.
Has anyone independently verified the 46% speed claim?
Not yet, based on available coverage. The benchmark was conducted by Keywords Studios, a studio contracted by Epic Games, and no independent outlet has published its own reproduced test results as of October 15, 2026.
What did Martin Keely say about the launcher’s performance?
Keely, vice president and general manager of the Epic Games Store, told IGN that “the library loads 30 times faster” and told GamesIndustry.biz that “the startup is twice as fast,” both comparisons measured against Epic’s previous launcher rather than Steam.
Should I switch to the new launcher now?
Tom’s Hardware specifically cautioned that the Preview build is not recommended for competitive players, since it is an early-stage release that is still missing features found in the stable client.
Did Steam or Valve respond to Epic’s claims?
No public response from Valve or a named Valve representative appears in the coverage reviewed as of October 15, 2026.
Does this affect pricing or revenue splits on the Epic Games Store?
No. The announcement is limited to the launcher’s client-side performance and platform support; no pricing or revenue-split changes have been reported alongside the release.
