Cuphead is a 2017 run and gun action game developed and published by Studio MDHR, built around a world-map campaign of tightly staged boss fights with a smaller set of run and gun levels. It launched on PC and Xbox in late 2017 and later expanded to additional console platforms, with a major DLC expansion arriving in 2022. The game’s identity matters for speedrunning because it is fundamentally a sequence of short, high-intensity encounters where mistakes are punished quickly, but attempts can be restarted just as quickly.
Cuphead speedruns tend to thrive on three evergreen ingredients: readable patterns, demanding execution, and meaningful decision-making. Bosses are built from distinct phases and attacks that can be learned, but still vary enough to keep runs from becoming rote. The controls are precise, the movement options are expressive, and the game’s combat systems reward aggressive play. In practice, that means runners are always balancing survivability against speed, choosing when to play safe and when to push damage.
A typical full-game run is an exercise in reducing “dead time” between victories. Runners optimize routing across the overworld, minimize menu friction, and make loadout choices that shorten fights rather than merely make them easier. Inside encounters, the core goal is to end phases as early as possible through efficient damage, consistent positioning, and smart meter usage. Parrying to build card meter, converting meter into EX attacks or Supers, and maintaining steady output while avoiding knockbacks and lost cycles becomes the heartbeat of high-level play.
What makes Cuphead distinct from many platformer-centric speedgames is that its fastest progress is usually earned in combat, not traversal. The best runs are less about one long continuous movement line and more about mastering dozens of micro-battles, each with its own optimal rhythm. That boss-first structure also makes the game unusually modular for practice and iteration, which helps explain why its speedrunning culture supports a wide spectrum of approaches, from execution-heavy consistency to strategies that lean harder on routing and menu proficiency.
Cuphead arrived in 2017 on Windows and Xbox One, with its audience widening later through additional platform releases. That rollout mattered for speedrunning because the earliest full game runners were often learning routes in parallel across different hardware and storefront builds, then comparing notes about what was consistent and what was version-specific. Early record-keeping followed the modern pattern for big indie releases: runs and route discussion circulated through uploaded videos and streams, but the scene’s public “ledger” quickly centered on speedrun.com, where categories, rules, and moderation could live in one place and be revised as standards hardened.
The first “standard goal” for full game play settled into a simple, readable premise: complete the required progression and finish by beating the Devil, with the main optimization problem being boss order and boss execution rather than exploratory movement. That clarity helped Cuphead develop multiple parallel identities at once. Full game categories could emphasize the cleanest path to the ending, while Individual Levels leaned on the game’s built-in scorecard timer so that single-stage performances could be compared across platforms with fewer variables. In practice, timing norms formed around a consistent start point on file select, with endpoints defined by category requirements and typically concluding at the Devil once those requirements were met.
As the game received updates, the community’s early history became tied to version policy. PC runners on storefronts like Steam and GOG could revert to older builds, while console players generally ran the most recent version available, which pushed organizers toward formal version splits to keep comparisons fair. That same era also established the loadless standard for many full game boards, reflecting a straightforward problem Cuphead exposed early: loading behavior varies by machine and can distort real time comparisons, so removing loads became a pragmatic way to keep the leaderboard focused on play rather than hardware.
As Cuphead’s speedrunning scene matured, discussion settled into a familiar two-layer structure: fast, day-to-day collaboration in real-time chat, and a stable public archive for rules and record-keeping. The community’s official “front page” for categories, leaderboards, and long-term standards is speedrun.com, where Cuphead’s boards centralize full-game categories, IL boards, and version groupings in one place. Alongside that public record, runners have long used a dedicated Discord server as the everyday workshop for routing talk, quick questions, and consistency troubleshooting, with the server link shared directly through the game’s forum space.
Verification and moderation are handled through structured roles rather than informal consensus, and those roles are visible in how runs appear on the leaderboard. A submitted run is reviewed and then marked “Verified” by a moderator or verifier, with the run page explicitly showing who submitted it and who verified it, along with key context like version and whether the time is loadless. The moderation framework itself distinguishes between moderators and verifiers, reflecting the practical reality of a busy board: runners can submit, but a separate group is responsible for confirming legality and maintaining consistency as categories evolve.
Knowledge preservation in Cuphead speedrunning tends to be a blend of living community memory and durable references. The game’s speedrun.com hub points people toward structured “Guides” and “Resources,” which act as a public bookshelf for routing notes and community tooling. On the tooling side, one of the most practical anchors is the community-maintained autosplitter and load remover hosted on GitHub, which standardizes how runners start and end timing and how they break a run into comparable segments. Together, that ecosystem lets Cuphead runners iterate quickly in chat while keeping the game’s rules, timing conventions, and reference materials discoverable for the next wave of players.
On speedrun.com, Cuphead’s leaderboards are organized around a few major “rulesets” that function like parallel eras of the game: a modern base-game board (“Version 1.1+”), a “Legacy Version” board for older builds, a “No Major Glitches” board, and separate boards for the DLC and for co-op. This structure matters because the game’s available tricks and consistency can change across versions, and because the DLC introduces new content and options that the community prefers to keep comparable within their own lanes.
Within the base-game “Version 1.1+” ruleset, the core full-game categories reflect different definitions of completion and different routing priorities: Any% as the fastest valid finish for that ruleset, All Flags as an expanded route that includes all Run and Gun stages alongside boss progress, Full Clear as the completionist category that aims to “do everything the game has to offer,” and Low% as a minimalist route where the spirit is to finish while collecting and doing as little as possible. The board also supports difficulty splits (Simple, Regular, Expert) as distinct competitive contexts rather than a single pooled ruleset.
Cuphead also has “named glitch” or “named ending” categories that exist specifically to keep comparisons fair and rules unambiguous. For example, the site lists an Any% (Menu Glitch) option, plus separate “Bad Ending” variants, rather than quietly allowing those approaches inside every category. This is one of the defining rule decisions for the game’s leaderboard identity: certain powerful strategies are treated as their own category families, while other boards (like “No Major Glitches”) exist to preserve a style of run that prioritizes clean execution and traditional routing over large skips.
Timing conventions are split by what is being measured. Full-game runs are commonly compared using “loadless” time when possible, especially on PC, to reduce hardware-dependent loading differences. Console runs, or PC runs without a load remover, are typically real time with loads. Individual Level runs lean on Cuphead’s built-in end-of-level timer, and the community also supports grade-focused IL boards where the goal is not just speed, but speed while meeting a target performance grade.
Submission rules and verification norms are built around making the run reviewable. The community guide used by runners emphasizes continuous capture, keeping key HUD elements visible, and keeping the loadless timer visible whenever load removal is used, so moderators can verify the time shown matches the run performed. In-game sound is also treated as a verification tool for competitive submissions, particularly for ILs. Across categories, the same “fair play” boundaries show up repeatedly: difficulty is not meant to be swapped mid-run, and third-party software beyond tightly limited utilities (like keybinders) is treated as disallowed.
Finally, version policy is not just a label, it is part of the rules. The scene distinguishes “Legacy” and “1.1+” as different competitive environments, and PC runners commonly use platform tooling (for example through Steam) to select specific builds that match the leaderboard’s version lane. That is why runs on LiveSplit and the load remover culture are not merely convenience features, they are part of how Cuphead speedrunning standardizes comparison across platforms and hardware.
Cuphead’s earliest route-building was shaped as much by which build you were playing as by raw execution. The scene quickly learned that some mechanics and exploits existed only in “legacy” versions, while later updates produced a more standardized environment for modern boards. Community documentation of version differences highlights several legacy-specific behaviors and tricks that changed how runners approached damage and phase management, including the presence of a weapon swap glitch and other fight or menu quirks that simply do not behave the same way on newer builds.
As the game’s patch landscape stabilized, the community’s competitive identity hardened around clear lines: some strategies became category-defining, and others became category-separating. The most famous example is the menu glitch, which is treated as its own category family on the main leaderboard rather than an assumed allowance everywhere. In the opposite direction, the “No Major Glitches” ruleset exists to preserve a cleaner style of run and, in practice, that has meant explicitly banning specific clipping tech that would otherwise dominate certain stages.
Within that framework, the biggest long-term evolution has been the shift from “survive and clear” to “control the fight.” Cuphead optimizations increasingly center on shortening boss cycles by choosing loadouts for damage, converting parries into meter at high reliability, and spending that meter in ways that end phases early instead of simply adding safety. In encounters where pattern variance matters, runners built tools and reference aids to turn chaos into repeatable decisions, including community resources that map out tricky pattern sets.
Tooling improvements helped make those gains comparable across machines. Load removal became part of the modern expectation for PC full-game runs, supported by a LiveSplit autosplitter/load remover designed for Cuphead that standardizes how runs are segmented and timed. Community discussion around retiming makes clear why this approach won out: the load remover detects loads through memory/state behavior that can begin and end on inconsistent visual frames, so manual “by eye” retimes do not reliably match what the plugin measures. The practical result is a culture where runners treat correct setup, including using game time when appropriate, as part of doing a legitimate, comparable run.
Practice and analysis tools also became a quiet pillar of consistency. The Cuphead community maintains an “Advanced Debug Mode” resource and related utilities that expose extra information useful for learning and refinement, including boss HP visibility and other run-relevant state. On the development side, the Cuphead debug mod project’s own release notes emphasize features geared toward training and controlled experimentation, such as expanded HUD options and additional control helpers, reflecting how the scene increasingly treats knowledge and repeatability as a form of speed in their own right.
One of the earliest turning points for Cuphead speedrunning was the moment its “public record” became more than a loose set of videos and personal PBs. As the boards on Speedrun.com filled out, the community gained a shared vocabulary for categories, timing expectations, and what counted as a comparable run. That shift did not just preserve faster times. It made Cuphead legible as a speedgame to outsiders, because the rules and leaderboards could be pointed to as a single, stable reference.
A second, category-defining milestone was the community’s decision to treat “menu glitch” as its own competitive identity rather than an assumed part of every run. The technique is powerful because it collapses repeated downtime, which means its presence reshapes the entire pacing of a route. Separating it into dedicated leaderboards also clarified what different audiences were watching: “fastest completion with this skip” versus “fastest completion without major route-breaking shortcuts.” That separation is visible in how the boards are structured and in the debate around whether the glitch belonged everywhere or in a distinct lane.
Version policy became another milestone because Cuphead is not “one game” from a routing standpoint. Legacy builds contain behaviors and damage or swap quirks that do not map cleanly onto later updates, and the scene responded by documenting differences and formalizing splits. That did more than protect fairness. It created a historical throughline, where older routes remain meaningful as artifacts of a specific rule environment rather than being erased by patch drift.
Cross-platform comparability also improved as timing and tooling matured. Loadless conventions and standardized split logic moved from being “nice to have” into a practical baseline for serious PC runs, because hardware-dependent loads can otherwise overwhelm small execution gains. The autosplitter and load remover ecosystem, paired with community guidance on how PC and console constraints differ, helped Cuphead become a game where two runs can be compared for play quality rather than for machine speed.
A quieter but enduring milestone was the emergence of a real documentation library. Once the scene had collected version notes, damage values, and route write-ups in public guides, optimization stopped being guarded knowledge and started being teachable. That changes what “progress” looks like in a mature leaderboard: breakthroughs become easier to validate, and new runners can climb faster because the fundamentals are archived rather than rediscovered.
Cuphead’s watershed visibility moments came through Games Done Quick showcases, where the game’s difficulty and spectacle translate well to a marathon audience. Runs like Summer Games Done Quick 2018 and Awesome Games Done Quick 2019 (including a prominently funded bonus-slot appearance) helped define Cuphead as a “main stage” speedgame and raised the baseline for what viewers expect: clean phases, confident aggression, and commentary that explains why riskier damage patterns pay off.
Finally, analysis and practice tooling became a milestone because it shifted improvement from intuition to measurement. Debug tools that expose boss HP and state information made it easier to test whether a strat is truly faster or just feels faster, and TAS tooling reinforced that idea by turning “possible” into something demonstrable and repeatable. Even when TAS is not the competitive format, its influence shows up in how communities talk about cycles, thresholds, and phase control.
SBDWolf – Version 1.1+ Any% (Regular, Loadless) – 27:01.500 – 18 days ago – PC / v1.1 – A modern benchmark for post-launch Cuphead Any%, where clean movement, consistent boss patterns, and fast transitions matter as much as raw execution.
Quincely0 – Version 1.1+ Any% (Regular, Loadless) – 27:12.570 – 3 months ago – PC / v1.1 – A top-era run that shows how refined “standard” 1.1 routes can get once a scene stabilizes around repeatable strats and tight segment goals.
Fintan0 – Legacy Version Any% (Regular, Loadless) – 22:46.290 – 1 year ago – PC / v1.0 – A defining Legacy-patch performance, highlighting why Cuphead’s speedrunning culture treats version splits as meaningful: different patches can produce genuinely different “fastest game.”
myekul – DLC + Base Game Solo Any% – 37:18.670 – 7 days ago – PC / v1.3 – A full-scope category snapshot where endurance, routing across two campaigns, and minimizing downtime between fights becomes the core optimization problem.
Sublime – Any% (Menu Glitch, Loadless) – 20:12.290 – 1 year ago – PC / v1.0 – The headline Menu Glitch category marker: it represents the “skip-heavy” branch of Cuphead running where category identity is defined by a single powerful technique and the rule set built around it.
hazelness – Any% (Menu Glitch, Loadless) – 20:55 – 3 years ago – PC / v1.0 – A major-era time close to the top that helps anchor the early competitive shape of the Menu Glitch leaderboard.
TheMexicanRunner (Mexico) – Any% (Menu Glitch, Loadless) – 23:15 – 8 years ago – PC / v1.0 – An early-record-era run from a high-visibility runner, useful historically because it captures what “fast Cuphead” looked like before later optimization layers tightened the category.
MarkinSws – DLC Solo Any% – 10:19.010 – 6 months ago – PC / v1.3 – A top DLC-only marker time, reflecting how quickly The Delicious Last Course developed its own streamlined route identity separate from base-game Any%.
GamerAttack27 – DLC Solo Any% – 10:19.300 – 15 days ago – PC / v1.3 – A near-tie at the top that shows the “seconds game” in mature categories, where micro-optimizations and consistency decide record swings.
GamerAttack27 – No Major Glitches Any% (Regular, Loadless) – 28:37.400 – 3 months ago – PC / v1.2 – A representative elite NMG time, useful for readers because it frames the “execution-first” side of Cuphead speedrunning without the biggest route-breaking glitches.
MarkinSws – No Major Glitches Any% (Regular, Loadless) – 28:45.410 – 1 year ago – PC / v1.2 – A historically meaningful top-tier NMG entry that helps show how close the ceiling can be even under stricter rule sets.
Studio MDHR. “Cuphead has a release date!” Accessed February 11, 2026. https://studiomdhr.com/cuphead-release-date/
Studio MDHR. “Cupdate – Patch Notes v1.1.3.” Accessed February 11, 2026. https://studiomdhr.com/cupdate-patch-notes-v1-1-3/
Steam Community. “Improve your DLC…with Patch 1.3.3 !!” July 19, 2022. Accessed February 11, 2026. https://steamcommunity.com/games/268910/announcements/detail/5241644120331776636
Steam Community. “Swap Your Title Screen…with patch 1.3.4!!” August 19, 2022. Accessed February 11, 2026. https://steamcommunity.com/games/268910/announcements/detail/3373779260018651815
Speedrun.com. “Cuphead.” Accessed February 11, 2026. https://www.speedrun.com/cuphead
Speedrun.com. “Cuphead Guides.” Accessed February 11, 2026. https://www.speedrun.com/cuphead/guides
Speedrun.com. “(For the Mods) With loads retiming.” Cuphead Forums. Accessed February 11, 2026. https://www.speedrun.com/cuphead/forums/h44cs
Speedrun.com. “How to Play cuphead in 1.1? & Auto Split.” Cuphead Forums. Accessed February 11, 2026. https://www.speedrun.com/cuphead/forums/atq1q
Calameo. “Guide To Cuphead Speedrunning V2 0.” Accessed February 11, 2026. https://www.calameo.com/books/007537711196060043c41
GitHub. ShootMe. “LiveSplit.Cuphead: Autosplitter for the game Cuphead.” Repository. Accessed February 11, 2026. https://github.com/ShootMe/LiveSplit.Cuphead
GitHub. DemoJameson. “Cuphead.DebugMod (Releases).” Repository. Accessed February 11, 2026. https://github.com/DemoJameson/Cuphead.DebugMod/releases
GitHub. DemoJameson. “Cuphead.TAS.” Repository. Accessed February 11, 2026. https://github.com/DemoJameson/Cuphead.TAS
Games Done Quick. “Cuphead by TheMexicanRunner in 48:42 – AGDQ2019.” YouTube video. Accessed February 11, 2026. https://www.youtube.com/watch?v=7AzL9jQq_tQ
Games Done Quick. “Cuphead by SBDWolf and Kirthar in 30:56 – Summer Games Done Quick 2025.” YouTube video. Accessed February 11, 2026. https://www.youtube.com/watch?v=fhgWhRRbRZ4
Games Done Quick. “Cuphead by GeneralAndrews in 30:12 – Games Done Quick.” YouTube video. Accessed February 11, 2026. https://www.youtube.com/watch?v=D4sbBFNO7lY
GDQ VODs. “GDQ VODs.” Accessed February 11, 2026. https://gdqvods.com/