The pandemic permanently shifted tabletop play online. By 2026, the majority of TTRPG sessions documented in the major communities run on a virtual tabletop (VTT) rather than around a physical table. Two platforms dominate this space. Roll20, the browser-based veteran from 2012, still leads in raw user count because of its zero-install accessibility and integration with the major publishers. Foundry VTT, the buy-once self-hosted alternative from 2020, has steadily eaten into Roll20’s GM share because of its deeper automation and better long-term economics. The choice between them is one of the most consequential decisions a new online GM makes. This is a practical look at how the two compare in 2026.
The two platforms at a glance
| Feature | Roll20 | Foundry VTT |
|---|---|---|
| Pricing | Free tier + $5/mo Plus + $10/mo Pro | $50 one-time, GM only |
| Hosting | Cloud-hosted (Roll20’s servers) | Self-hosted or third-party hosted |
| Player install | None, browser only | None, browser only |
| GM setup time | Under 30 minutes | 2-4 hours initial |
| Automation depth | Medium | Deep, system-dependent |
| Best system support | D&D 5e, generic | Pathfinder 2e, D&D 5e, many others |
| Marketplace | Roll20 native + D&D Beyond | DriveThruRPG, Paizo, community modules |
These numbers reflect the platforms as they ship in 2026. Both update frequently and the gap on specific features can change session to session. The structural differences (pricing model, hosting, automation philosophy) are stable.
Roll20: the accessible default
Roll20 became the dominant VTT because it removed friction. A player joining a Roll20 game in 2026 clicks a link, makes a free account, and is at the table in two minutes. No download. No configuration. No system requirements beyond a working browser. For groups where the players are not technical and the GM wants to minimize support questions, this still matters enormously.
Roll20’s official content marketplace is also a real advantage for casual play. Wizards of the Coast modules, Paizo Adventure Paths, and dozens of indie publishers sell pre-built Roll20 versions of their books that import with all the maps, monsters, handouts, and tokens already configured. A GM can buy a Dragonlance: Shadow of the Dragon Queen module for twenty-five dollars and be running session one within an hour. This is genuinely faster than Foundry for the first session of a published campaign.
The downsides of Roll20 are real and accumulate over a long campaign. The free tier limits dynamic lighting, advanced fog of war, animated tokens, and storage. The Plus tier at five dollars a month adds most of those. The Pro tier at ten dollars a month adds API scripting and a few advanced features. A GM running weekly for a year on the Plus tier pays sixty dollars, which is more than Foundry’s one-time fifty-dollar fee for less capability. The math gets worse for longer campaigns.
The other downside is automation depth. Roll20’s character sheets are good for 5e and acceptable for most other systems. They are not as deep as Foundry’s system modules. A complex Pathfinder 2e build with multiple class feats, archetype feats, and skill feats can take an hour to assemble on a Roll20 sheet and ten to twenty minutes on Foundry. For one-shots this rarely matters. For long campaigns with leveling-up sessions, the time cost adds up.
Foundry VTT: the power user choice
Foundry took the opposite approach: charge the GM once, give them a self-hosted application, and let them install community modules to customize the experience. The cost is fifty dollars total for the GM. Players join free through a browser link, the same as Roll20. The economics favor Foundry within a few months for any regular campaign.
The technical advantage is automation. Foundry’s system modules (DnD5e, PF2e, Cthulhu, Lancer, and dozens more) are maintained by communities of developers who care about deep rule support. The PF2e system in particular is maintained with Paizo’s blessing and automates conditions, three-action economy, attack-of-opportunity triggers, and feat interactions to a degree Roll20 has never matched. The 5e system similarly automates spell effects, weapon mastery, exhaustion tracking, and concentration checks well enough that a GM rarely has to look up a rule mid-combat.
The module ecosystem is the second technical advantage. Foundry’s open architecture lets developers ship community modules that add fog of war animations, automated lighting, sound packs, encounter trackers, integrated music players, and dozens of other features. A typical experienced Foundry GM runs twenty to forty modules in a campaign. Setting these up the first time takes an evening. Maintaining them as Foundry updates can require occasional troubleshooting, which is the real cost of the customization.
The downside is hosting. Foundry is a self-hosted application by default. The GM either runs it on a home PC (which works but requires port forwarding or a tunneling service for player access), installs it on a home server, or pays a third party (Forge, Molten Hosting, Oracle Cloud Free Tier) to host it. Most GMs eventually settle on hosted service for reliability. The hosted options range from free to about ten dollars a month and are still cheaper than Roll20 Plus over a year.
What each platform is best for
A few specific scenarios point clearly toward one platform.
For a one-shot at a convention or a one-time game with strangers, Roll20 is the better choice. The lower setup time and the shareable link model fit casual play. The free tier is sufficient for a four-hour session.
For a weekly long-running D&D 5e campaign, either platform works. Roll20 is fine for narrative-heavy campaigns where the players do not care about automation. Foundry is better for combat-heavy campaigns where the GM appreciates the deeper rule support.
For a Pathfinder 2e campaign, Foundry is clearly the better choice. The PF2e system module is the best automation any VTT delivers for any system. Running Pathfinder 2e on Roll20 is possible but loses much of what makes the system mechanically interesting.
For a Call of Cthulhu campaign, either platform works because the system is light on combat automation. Roll20 has more pre-built published Cthulhu scenarios. Foundry has better atmospheric tools (lighting, fog, sound) which match the genre.
For a GM who is not technical and dreads troubleshooting, Roll20 is the safer choice. For a GM who enjoys customization and is willing to spend an evening setting up modules, Foundry pays back the investment many times over.
The migration question
Many GMs who started on Roll20 in 2020-2022 have since migrated to Foundry. The migration path is straightforward for character sheets in 5e and PF2e. Maps and tokens can be imported directly. Campaign notes and handouts usually have to be rebuilt manually. Most GMs estimate ten to twenty hours of work to migrate a long campaign cleanly.
The decision to migrate usually comes when one of two things happens. Either the Roll20 subscription becomes a recurring annoyance and the GM realizes Foundry would have paid for itself months ago, or a specific mechanical limitation in Roll20 forces a workaround that Foundry would handle natively. Either trigger is reasonable. Most GMs who migrate report they should have done it sooner.
See our look at running better online sessions for tips that apply to both platforms, and our comparison of TTRPG starter sets for which physical product to buy before going digital.
Frequently asked questions
Is Foundry VTT worth the fifty dollar one-time cost?+
For any GM running more than a one-shot, yes. Foundry's pricing model charges the GM once and lets unlimited players join free. Across a year-long campaign that runs eighty sessions, the cost works out to under a dollar per session. Roll20's free tier is capped on map sizes and table features, and the Plus and Pro subscriptions add up to more than Foundry's one-time fee within four to six months. The Foundry cost is recovered quickly for anyone running regularly.
How steep is the Foundry learning curve really?+
Steeper than Roll20 in the first week, gentler after that. Foundry expects the GM to install the application, configure a world, install system and module support, and learn the macro system to unlock the platform's full power. The initial setup runs two to four hours for a new GM. After that, running sessions is faster than Roll20 because the automation handles more of the rules. Most GMs who switch from Roll20 to Foundry report the first session feels harder and the tenth feels noticeably faster.
Does Roll20 still make sense in 2026?+
Yes for two cases: GMs who run irregular one-shots and groups whose players are already comfortable on the platform. Roll20's strengths are the zero-install browser-based access and the marketplace integration with D&D Beyond and Paizo content. A GM who runs a one-shot every two months at a convention or for a casual friend group does not benefit from Foundry's deeper automation. A GM running weekly campaigns probably does.
Which platform handles Pathfinder 2e better?+
Foundry, by a meaningful margin. The Pathfinder 2e system module for Foundry is maintained by Paizo and the community at a higher quality level than the Roll20 sheet. Three-action economy, condition tracking, and the dense feat chains all automate cleanly in Foundry. Roll20's PF2e sheet is functional but requires more manual rule lookups. For 5e the gap is smaller because both platforms automate 5e well, though Foundry's Tidy 5e and DnD5e module ecosystem still edges ahead.
Can I use Foundry without paying for hosting?+
Yes, in three ways. The simplest is to run Foundry locally on the GM's home computer and have players connect through the GM's port forwarding or a tunneling service like ngrok. The second is to install Foundry on a free or cheap home server. The third is to use a hosted service like Forge, Molten Hosting, or Oracle Cloud Free Tier, which costs zero to ten dollars a month depending on the option. Most GMs eventually move to a hosted setup because port forwarding from a home PC is fragile.
Priya Sharma
Priya Sharma writes for The Tested Hub.