DLSS5 Autopilot: More Games, Lower Cost [2026]

DLSS 5 Autopilot: Lower Cost Neural Rendering for More Games Per VideoCardz Sept 7 [2026 Full Guide]

By Indie Kings | September 8, 2026 | Updated September 8, 2026 | 18 min read

Photo: Pexels free use.

On September 7, 2026, VideoCardz reported that the community tool DLSS5-Autopilot can now install and configure Neural Upstream, a separate mod by matiasLombo that runs DLSS 5 Neural Rendering at lower internal resolution before DLSS Super Resolution. That single pipeline change cuts the performance cost of modded neural rendering in a major way. This report is single source attributed to that VideoCardz September 7 article by WhyCry, plus the linked DLSS Autopilot GitHub and Reddit thread it cites. No other outlet is used for the core claim. In this full guide we explain what Autopilot does, what Neural Upstream changes versus the full regular route, measured fps from Assassin's Creed Shadows plus Stalker 2 plus GTA V Enhanced plus Bright Memory Infinite, VideoCardz own tests in Company of Heroes 3 and Need for Speed Hot Pursuit Remastered, who benefits most including RTX 40 and RTX 30 laptop owners, setup logic, visual differences, anti cheat risk, and what to watch next. For live GPU plus game bundle prices see our Bundle Tracker.

Table of Contents

• 1. Single Source Note: What VideoCardz Reported Sept 7
• 2. What DLSS5-Autopilot Does in Plain Terms
• 3. Full Route vs Neural Upstream: What Changes
• 4. FPS Evidence: 63 vs 31 vs 50 and Stalker 2 Claims
• 5. Game by Game: Tested and User Reported Titles
• 6. Who Benefits Most: RTX 50, 40, 30 and Laptops
• 7. Visual Result: What Neural Rendering Adds
• 8. Setup Flow and Performance Tuning
• 9. Risk Check: Anti Cheat, Online Games, Stability
• 10. What To Watch Next and Verdict

1. Single Source Note: What VideoCardz Reported Sept 7

VideoCardz published the Autopilot report on September 7, 2026 at 17:12 GMT, authored by WhyCry. The headline states DLSS 5 Autopilot brings Neural Rendering to more games with lower performance cost. The body makes three factual claims we reuse here with attribution. First, DLSS5-Autopilot automates several community methods for testing Neural Rendering in games without official support. Second, it can now use Neural Upstream to process Neural Rendering before DLSS Super Resolution, changing where the model runs in the pipeline. Third, results vary heavily between games. VideoCardz cites DLSS Autopilot on GitHub, a Reddit thread on r slash DLSS, and a BratorNews post from September 6 about up to 50 percent faster results in Stalker 2 on RTX 4080. That is the full source chain for this guide. No Nvidia statement, no second press outlet, and no independent lab bench is added for the core Autopilot plus Upstream claim. Where we give context on DLSS 5 launch timing, such as the September 3 NBA 2K27 RTX 50 launch and the RTX 40 fall plan, that context is framed as background already covered in our prior guides, not as new reporting. The reason for strict single source attribution is simple. Community mod performance moves fast, numbers differ by scene and driver, and mixing outlets creates false precision. One source, one date, one tool version logic keeps expectations honest. Readers should treat every fps figure below as scene specific evidence from that report, not as a universal guarantee. Tool updates after September 7 can shift results. Check the GitHub releases page before you test. See current DLSS capable GPU bundle value in our Bundle Tracker.

What VideoCardz actually tested themselves matters for trust. They tried Autopilot with Company of Heroes 3 and Need for Speed Hot Pursuit Remastered. Need for Speed stayed playable near 30 fps on an RTX 3070 Laptop GPU with a visible effect. Company of Heroes 3 showed color and texture shifts but stayed too slow for comfortable play in their test without upscaling enabled. They note performance could still improve with upscaling on. They also warn against testing in multiplayer or anti cheat protected titles because the tools rely on DLL injection, ReShade, and rendering hooks that can trigger integrity systems. Our guide respects those limits and does not recommend online testing. Every section below traces back to that September 7 page. If a detail is not on that page, we label it as practical guidance, not as reported fact.

2. What DLSS5-Autopilot Does in Plain Terms

DLSS5-Autopilot is a Windows community tool that scans a game install, detects architecture and graphics API, picks a compatible community route, fetches components, and writes the configuration so DLSS 5 Neural Rendering can load in titles that never shipped with it. Think of it as an installer plus configurator, not as a new neural model. It does not contain Nvidia binaries in the repo itself. It pulls needed files at runtime from original publishers. The GitHub project states it holds installer logic only, with no game files and no third party redistributables inside. That design matters for safety and updates. When Nvidia or community authors revise a DLL, Autopilot can fetch the newer file without a full tool rebuild. For users, the flow is one executable with no formal install step. Run, pick a game, let it analyze, apply the route, launch and verify. Prior mod paths required manual DLL swaps, ReShade add on wiring, OptiScaler forks, registry style profile tweaks, and long forum threads. Autopilot compresses that into guided clicks. It also matters because DLSS 5 official support remains narrow. Nvidia launched DLSS 5 on September 3 in NBA 2K27 for RTX 50 GPUs, with RTX 40 support planned for fall and older cards unconfirmed. The mod scene filled the gap with ReShade add ons and an OptiScaler fork, plus retargeted runtimes that let RTX 40, 30, and 20 attempt the workload. Autopilot sits on top of that scene as the friendly front end. It does not make every game compatible. API, engine hooks, HDR chain, overlay software, and driver version all affect success. But for supported single player DirectX titles, it lowers the skill floor from modder level to enthusiast level. Track GPUs that can run these mods in our Bundle Tracker.

The practical value is testing breadth. Before Autopilot, only a handful of users tried neural rendering outside official titles because setup took hours. After Autopilot, VideoCardz could test two extra games in days, and forum users reported Cyberpunk 2077 and Stalker 2 plus GTA V Enhanced and Bright Memory Infinite results in the same week. That speed is why the September 7 report exists. Automation turned isolated experiments into comparable data. Users should still keep expectations measured. Automation does not fix engine incompatibility, shader conflicts, or weak GPUs. It only removes setup friction. The heavy lifting still happens on Tensor cores and in VRAM, which is why pipeline order discussed next matters so much for fps.

3. Full Route vs Neural Upstream: What Changes

The regular modded route runs Neural Rendering after DLSS Super Resolution. That means the neural network processes pixels at final output resolution. At 4K output, that is a full 3840 by 2160 frame worth of neural work every frame, plus masks, plus intensity and grading controls. Image quality can be strong because the model sees the sharpest input, but cost is high because pixel count is highest at that stage. Early leaked DLL tests showed exactly that problem, with large base cost before any frame generation synthesis, sometimes cutting fps in half or worse on cards below RTX 5090 class. Neural Upstream flips the order. It applies Neural Rendering at the lower internal render resolution, then sends that result through DLSS Super Resolution to reach output resolution. If you run DLSS Performance mode at 4K, internal resolution is near 1080p class, roughly one quarter the pixels of 4K. The neural model therefore processes far fewer pixels, then the upscaler reconstructs detail to 4K. Cost falls sharply while most of the relight plus material plus tone character remains. That is the core change per VideoCardz. Same model family, earlier in the pipe, lower pixel load, then upscale. It is not a new Nvidia official mode. It is a community pipeline reorder by matiasLombo that Autopilot can now install and configure automatically. For buyers, the analogy is simple. Full route is like color grading a finished 4K master. Upstream is like grading a 1080p proxy then upscaling. The second path is far cheaper and usually close enough visually, especially in motion. Check GPU plus monitor bundle math in our Bundle Tracker.

What does not change is also important. Upstream does not add official support, does not bypass hardware limits, and does not remove VRAM need. The model buffers, masks, and per object controls still consume memory. A 4K Quality chain with neural passes still prefers 12GB plus, ideally 16GB plus for long sessions with high textures. Upstream also does not fix artifacts by magic. Running the model at lower resolution can soften fine material response on skin, hair, fabric, and thin geometry, then the upscaler must reconstruct that softened signal. In some scenes that looks cleaner because noise is suppressed early. In other scenes it looks slightly less precise than full resolution neural work. VideoCardz notes results vary heavily between games, which matches this tradeoff. Engine lighting, tone mapping order, and mask quality decide whether Upstream looks near identical or visibly different. Users should compare screenshots at same time of day and same camera, not just fps.

4. FPS Evidence: 63 vs 31 vs 50 and Stalker 2 Claims

The clearest number in the VideoCardz September 7 report is an RTX 4080 test in Assassins Creed Shadows at 4K with DLSS Performance. Baseline without Neural Rendering sits near 63 fps. With the regular full route it falls to 31 fps. With Neural Upstream it recovers to roughly 50 fps. That is a 51 percent drop for full route versus baseline, but only a 21 percent drop for Upstream versus baseline. Put another way, Upstream restores about 19 fps of the 32 fps lost to full neural work in that scene, keeping 79 percent of original speed while keeping the neural look. That single test explains the headline lower performance cost. It does not prove every game behaves the same, but it proves pipeline order can matter more than GPU overclock in modded DLSS 5. A 10 percent core overclock cannot recover 19 fps at 4K. A pipeline reorder can. For RTX 4080 owners, that is the difference between a slideshow style 31 fps that needs frame generation to feel smooth and a near 50 fps base that needs only light synthesis or a modest settings tweak to lock 60. For context, 63 fps at 4K Performance on RTX 4080 in a dense open world is already a tuned result. Adding any neural pass will cost frames. The question is how many, and Upstream answers with far fewer. Compare RTX 4080 street plus PSU cost in our Bundle Tracker.

The second claim comes via BratorNews on September 6, cited by VideoCardz. Modders fixed the performance cost before Nvidia, with up to 50 percent faster Stalker 2 on RTX 4080 when using Upstream versus the prior mod path. Treat up to 50 percent as a best case scene delta, not an average. Stalker 2 has heavy foliage, volumetric light, and dense material variation that stresses neural passes at full res. Moving that work to internal res helps disproportionately in foliage plus fog scenes. In indoor or low complexity scenes the gap will be smaller. Still, a 50 percent uplift in the hardest scenes is exactly where playability is decided. One percent lows rise, traversal hitches fall, and frame generation multipliers have a higher base to work from. A 30 fps base times 2x synthesis equals 60 displayed. A 45 fps base times 2x equals 90 displayed. Same synthesis cost, different feel. That math is why Upstream plus modest frame generation is the practical combo for RTX 40 cards until official RTX 40 support lands this fall. Wait for your own scene logs before declaring victory. Record average plus one percent lows over a 60 second traversal loop, not a static vista.

5. Game by Game: Tested and User Reported Titles

VideoCardz lists two test tiers. Lab tested by VideoCardz: Company of Heroes 3 and Need for Speed Hot Pursuit Remastered. Community tested with Upstream: Assassins Creed Shadows, GTA V Enhanced, Bright Memory Infinite, plus user reports in Cyberpunk 2077 and Stalker 2. Each behaves differently because engines, lighting, and mask hooks differ. Assassins Creed Shadows shows the biggest documented win, 63 to 31 to 50 fps on RTX 4080 at 4K Performance, with lighting and material response surviving the Upstream move. That makes sense for an open world with strong sun plus skin plus cloth variation where neural relight is visible but does not need full res to read well. GTA V Enhanced is an older engine with enhanced lighting paths, where Upstream reportedly runs with lower cost and stable output because base pixel shading is simpler and masks are cleaner. Bright Memory Infinite is a linear shooter with controlled scenes, another friendly case for Upstream because camera paths and light rigs are predictable. Cyberpunk 2077 and Stalker 2 user reports are promising but more variable. Both push ray tracing, dense crowds or foliage, and complex transparencies that stress mask quality. Expect scene to scene variance of 10 to 20 fps even on the same GPU. Test a city walk, a night neon loop, and a combat burst before judging. See capture card plus storage bundles for clip testing in our Bundle Tracker.

VideoCardz own two added titles tell a cautionary tale. Need for Speed Hot Pursuit Remastered produced the better result and remained playable near 30 fps on RTX 3070 Laptop GPU, with the neural effect easier to spot. That is notable because RTX 3070 Laptop is far below RTX 4080 desktop, yet Upstream plus modest settings kept an older arcade racer playable with visible relight. Car paint, road specular, and sky tone are exactly the broad material signals that survive lower res neural work well. Company of Heroes 3 mostly showed different color grading and changes to some textures, while performance stayed too low for comfortable play in their no upscaling test. Strategy maps with many small units plus UI overlays plus CPU bound simulation do not flatter a GPU neural pass. They also did not enable upscaling in those tests, so headroom remains untapped. Enable DLSS Performance or Balanced, retest, then judge. The lesson across all seven titles is consistent with the header warning. Results vary heavily between games. Engine first, GPU second, settings third. Do not assume Shadows numbers transfer to Heroes.

6. Who Benefits Most: RTX 50, 40, 30 and Laptops

RTX 50 owners benefit from choice. They already have official DLSS 5 in NBA 2K27 and a forward path as more titles add tuned support. Autopilot plus Upstream gives them a lab for every other game in the library today, with lower cost than early full route mods. An RTX 5070 or 5070 Ti at 1440p can test Upstream with Balanced upscaling and keep high refresh headroom. An RTX 5080 at 4K can test Upstream with Performance upscaling and keep near 60 base in many single player titles before frame generation. An RTX 5090 can brute force full route if desired, but even halo owners may prefer Upstream for lower latency and cooler operation. VRAM is less of a wall on 50 series with 16GB plus, yet Upstream still helps because Tensor load, not just capacity, decides fps. Lower pixel load means lower Tensor occupancy, which leaves headroom for frame generation and Ray Reconstruction. If you own RTX 50, use official where available, Upstream mod where not, and keep drivers current. Log one percent lows because neural passes affect lows more than averages. Find RTX 50 bundle deals in our Bundle Tracker.

RTX 40 and RTX 30 owners benefit from access. Nvidia has said RTX 40 support is planned for fall without a firm date, and has not confirmed older cards. Mods already let RTX 40, 30, and even 20 attempt neural rendering, with heavy cost on the old full path. Upstream is what makes that access playable rather than theoretical. RTX 4080 at 4K is the proof point at 50 fps. RTX 4070 class at 1440p should scale similarly with Balanced mode. RTX 3070 Laptop near 30 fps in Need for Speed shows the floor is real for older mobile chips when the game is light and upscaling is used wisely. Desktop RTX 3080 10GB and 3090 24GB split on VRAM. The 3090 absorbs buffers better for long sessions, while the 3080 needs tighter textures to avoid paging that looks like neural slowness but is really memory spill. RTX 20 owners should treat Upstream as experimental, 1080p plus Performance upscale only, with modest expectations. The ladder is clear. Newer Tensor cores plus more VRAM plus lighter target resolution equals better Upstream results. Older laptop plus 4K plus full route equals disappointment. Match ambition to hardware.

7. Visual Result: What Neural Rendering Adds

DLSS 5 Neural Rendering is not another sharpness slider. Per Nvidia framing reused across our guides, it takes color plus motion vector data and applies a trained model that improves lighting and material response while staying anchored to the 3D scene. In practice that means skin subsurface tone, hair light interaction, fabric sheen, car paint specular, road wet response, and sky plus fog grading shift toward a more photo style render. VideoCardz comparison shots for Need for Speed show exactly that accessible change. Road tone, car body reflections, and sky grade read differently with the pass on, even at modest resolution. Company of Heroes 3 shows a subtler shift focused on color grading plus some textures, which matches a top down strategy view where fine material detail is small on screen. Neither result is a simple resolution boost. It is a relight plus regrade that rides on top of the rendered frame. That is why fps cost exists even before upscaling. The model runs every frame, deterministically, to avoid flicker. Determinism costs compute. Upstream keeps determinism but runs it on fewer pixels, which is why most of the look survives while fps recovers. For screenshot hunters, full route may still win pixel peeping contests on hair strands and thin nets. For players in motion, Upstream keeps 90 percent of the character for far less cost. Judge in motion at your play distance, not at 400 percent zoom. See 4K display bundles in our Bundle Tracker.

Mask and control behavior also shapes the look. DLSS 5 gives developers intensity, color grading, and per object masking controls through Streamline. Community routes approximate those controls without per title tuning, so some scenes over apply the look while others under apply it. That explains heavy game to game variance. A game with clean object IDs and stable motion vectors gives the model good guidance. A game with transparencies, particles, UI heavy overlays, or modded shaders gives noisy guidance. Upstream can actually help noisy cases by smoothing input before the neural pass, but it can also soften intended fine detail. Practical rule: if faces look waxy or text shimmers, lower neural intensity if exposed, or step upscale from Performance to Quality to feed cleaner input. If the scene looks flat with the pass off and rich with it on, Upstream is doing its job. Save before and after captures with identical weather plus time plus camera to avoid fooling yourself.

8. Setup Flow and Performance Tuning

Setup with Autopilot follows a repeatable loop. Update to a recent GeForce driver that already contains DLSS 5 profile traces, clean install if switching branches. Download the latest Autopilot release from GitHub, keep it in its own folder, allow it to fetch components at runtime. Run as admin only if the game folder needs permission, otherwise standard user is safer. Point Autopilot at the game exe, let it detect API and architecture, pick the Upstream route where offered, apply, then launch the game from its normal launcher. Inside the game, set display mode to exclusive fullscreen or borderless as Autopilot docs suggest for that title, enable DLSS Super Resolution first, pick Performance at 4K or Balanced at 1440p for first test, turn off other injectors and overlays. Test a 60 second traversal loop with CapFrameX or Nvidia overlay metrics, log average plus one percent lows plus VRAM. If stable, add frame generation at 2x, retest, then consider 3x or 4x only if base stays above 45 fps. Keep notes per game because one global setting never fits all engines. Store games on NVMe SSD because streaming hitches mimic neural cost. Find SSD plus PSU bundles in our Bundle Tracker.

Tuning order matters more than raw sliders. First fix resolution and upscale ratio, because pixel load dominates neural cost. Second fix textures to fit VRAM, because paging destroys lows. Third fix ray tracing level, because RT plus neural stacks cost multiplicatively. Fourth add frame generation, because synthesis multiplies whatever base you earned. Fifth adjust neural intensity or grading if exposed. Example path for RTX 4080 at 4K: DLSS Performance, textures High not Ultra plus mods off, ray tracing High not Overdrive, Upstream on, verify near 50 fps in Shadows style scenes, then enable 2x synthesis for near 100 displayed. Example path for RTX 3070 Laptop at 1080p or 1440p: DLSS Balanced or Performance, textures Medium to High, RT off or Low, Upstream on, cap to 30 or 60 for stability. If fps still misses, drop output to 1440p before blaming the mod. Output resolution is the heaviest lever. Keep a clean restore point. Back up original DLLs, know how to verify game files, and remove hooks before playing anything with anti cheat.

9. Risk Check: Anti Cheat, Online Games, Stability

VideoCardz states the risk plainly. These tools rely on DLL injection, ReShade, and other rendering hooks that may trigger anti cheat or integrity systems. Do not test DLSS 5 mods in multiplayer games or anything protected by anti cheat. That warning covers Company of Heroes 3 and Need for Speed titles they tested, both of which have online functionality, and it covers every live service title readers might be tempted to try. Anti cheat systems flag unknown DLLs, altered render chains, and overlay hooks even when intent is visual only. A ban can be automatic and non reversible. The safe list is short. Single player only, offline mode where possible, anti cheat uninstalled or never installed, no ranked or co op session, no shared account with a main ranked profile. Even in single player, some games with online checks validate files on launch and may refuse to start or may overwrite modded DLLs. That is inconvenient but not dangerous. The dangerous case is kernel level anti cheat that reports on first injection. When in doubt, skip the title. No relight effect is worth a library ban. Keep mod testing on a separate Steam or publisher account if the platform allows family sharing boundaries that isolate risk, and read the anti cheat terms for each title before hooking. See single player game bundle deals in our Bundle Tracker.

Stability risk is lower but real. Modded neural passes can crash on driver update, on game patch day, on HDR toggle, on alt tab with overlays, or on specific weather plus particle combos. Symptoms include black screen on launch, flickering masks on thin geometry, waxy skin from over intensity, text shimmer at high synthesis multipliers, and sudden VRAM spikes in long sessions. Mitigation is methodical. Test one change at a time. Disable other mods, overlays, and capture hooks during first run. Verify game files after removing the mod to restore clean DLLs. Avoid leaked DLLs from random mirrors, use Autopilot fetched publisher files or known community releases with hashes. Keep a known good driver version noted per game, because new Game Ready branches can shift profile behavior. Back up saves before long sessions. If a title patches weekly, expect to reapply or wait for Autopilot update. Treat this as a test bench hobby, not as a set and forget driver feature, until Nvidia ships official support for your card and your game.

10. What To Watch Next and Verdict

Watch three signals after September 7. First, Autopilot release notes for broader Upstream coverage and per game presets. If Cyberpunk 2077 and Stalker 2 gain one click Upstream profiles with mask tuning, their variance will shrink and averages will rise. Second, Nvidia RTX 40 fall support timing plus driver profile changes. Official support could obsolete mods for many RTX 40 owners overnight, or it could validate Upstream style ordering if Nvidia adopts lower res neural work officially. Third, independent frame time tests from Digital Foundry style outlets comparing full versus Upstream versus off at matched scenes with one percent lows. Averages will look close in light scenes. Lows in heavy traversal will separate. Also watch mod unlocks for RTX 20, 30, and 40 frame generation interplay, plus any Nvidia blocks and immediate community workarounds already seen in early September. Memory supply and GPU street prices decide who can actually use these mods. As long as RTX 50 cards sell above MSRP, the practical audience for official DLSS 5 stays smaller than marketed, which keeps Autopilot relevant for older cards. Track street normalization in our Bundle Tracker.

Verdict from a single source view is narrow but useful. Autopilot plus Neural Upstream does not make DLSS 5 free, but it makes modded neural rendering playable in more games on more cards by moving work before upscaling. The Shadows 63 to 31 to 50 fps ladder and the Stalker 2 up to 50 percent claim point the same way. Pipeline order beats brute force for cost control. Who should try it now. Single player gamers on RTX 4080 class or better at 4K, RTX 4070 class at 1440p, and adventurous RTX 3070 laptop owners in light titles, all offline and with backups. Who should wait. Anyone on a ranked account, anyone on anti cheat protected live games, anyone who needs zero crash streaming stability, and RTX 40 owners who prefer to wait for official fall support. Single source attribution holds for the core claim. Everything else here is practical guidance to use that claim safely. Revisit after official RTX 40 numbers land, then compare official versus Upstream head to head at same scenes.

FAQ: 6 Answers

1. What did VideoCardz report on Sept 7 about Autopilot?
That DLSS5-Autopilot can now install and configure Neural Upstream, a mod by matiasLombo that runs Neural Rendering before DLSS Super Resolution at lower internal resolution, cutting cost. Tests include Shadows 63 base to 31 full to 50 Upstream on RTX 4080 at 4K Performance, plus GTA V Enhanced and Bright Memory Infinite, plus user reports in Cyberpunk 2077 and Stalker 2. VideoCardz own tests added Company of Heroes 3 and Need for Speed HP Remastered, with heavy game to game variance.

2. What changes vs the full regular modded route?
Full route runs the neural model after upscaling at output resolution, such as full 4K pixels, with high cost. Upstream runs it before upscaling at internal resolution, such as near 1080p class for 4K Performance mode, then upscales. Pixel load falls sharply, fps recovers most of the loss, and most relight character remains. VRAM need and mask quality still matter.

3. Who benefits most from Upstream?
RTX 4080 class at 4K, RTX 4070 class at 1440p, and older cards like RTX 3070 Laptop in light titles benefit most because Upstream turns slideshow full route fps into playable base fps. RTX 50 owners gain choice across unsupported games. RTX 40 owners gain a playable bridge until official fall support. RTX 30 and 20 owners should stay at lower output resolutions with modest expectations.

4. Which games work with Autopilot plus Upstream?
Per the Sept 7 report: Assassins Creed Shadows, GTA V Enhanced, Bright Memory Infinite, Cyberpunk 2077 and Stalker 2 via community reports, plus Need for Speed Hot Pursuit Remastered and Company of Heroes 3 via VideoCardz own tests. Results vary heavily. Shadows shows the clearest win. Need for Speed is playable on modest laptops. Heroes needs upscaling enabled and still may disappoint for strategy map loads.

5. Is it safe to use in online or anti cheat games?
No. VideoCardz warns these tools use DLL injection, ReShade, and rendering hooks that can trigger anti cheat or integrity systems. Do not test in multiplayer or protected titles. Use single player offline only, back up DLLs, verify files after removal, and keep mods off ranked accounts.

6. Should you use Upstream now or wait for official support?
Try Upstream now if you play single player offline, own a capable RTX card, accept test bench quirks, and want neural rendering beyond NBA 2K27 today. Wait if you need zero crash stability, play protected live games, or own RTX 40 and prefer official fall support. Either way, log average plus one percent lows over traversal loops and compare Upstream versus off at same scenes before deciding.

Written by Indie Kings, September 8, 2026. Single source: VideoCardz Sept 7, 2026 WhyCry Autopilot plus Neural Upstream report with linked GitHub and Reddit. For live GPU plus game bundle prices see Bundle Tracker.

Related: Bundle Tracker | VideoCardz Sept 7 Autopilot Report | DLSS5-Autopilot GitHub