MARS · MIRROR LEDGER · 2026
Mars Mirror List 2026: Every Verified Onion Address and Its Status
All five addresses we track for Mars, written out so you copy them by hand. No shortened links, no buttons that hide the string. Each row carries a status word, and this page is mostly about what those words really mean — for the clone-spotting and I2P-fallback angle see mars-darknet.com's mirror page, and for a role-based registry (primary/failover/i2p) see mars-market.vip's mirror list.
Checking, dead, and why you never see Online
Three words show up on this page, and only two of them ever appear next to a live address. Knowing which is which saves you from a clone that leans on a confident-looking badge.
Checking
A probe has been sent and we are waiting on the answer. It means “not yet confirmed,” not “safe.” You still verify before you open it.
Dead
The address stopped answering from every vantage we tried. We keep it on the page, struck through, so you can recognise it and walk past it.
No Online badge
We do not paint a permanent green light. A status that never changes is a decoration, and decorations are exactly what a phishing mirror copies first.
How a row moves from queued to a marker you can read
A status is not typed in by hand. It is the last stop in a short loop that runs, times out, and records what it saw.
How to read a Mars status row
Each row on the ledger below has three parts worth reading in a fixed order. First, the role — primary, mirror, or legacy — which tells you whether an address is meant to be your first stop or a fallback. Second, the address itself, which is the only part that matters once you are ready to connect and which you should copy in full rather than retype. Third, the marker, which is a probe result, not a guarantee — read it as "the last time we checked, from our vantage point, this is what happened," not as a live promise.
Why a Mars mirror can be Checking and still be genuine
A Tor circuit can fail for reasons that have nothing to do with the destination: a slow guard relay, a congested path, a directory lookup that timed out on our end rather than the service's. Checking captures all of that ambiguity honestly instead of guessing. If a mirror has read Checking for a long stretch, that is more informative than a single reading, but even then the fix is to re-test yourself rather than assume either an outage or a fabricated Online badge.
What to do when every mirror reads the same marker
If every row on this page reads Checking at once, the more likely explanation is a probing-side issue — a Tor bootstrap problem or a network change on our monitoring host — rather than every Mars mirror failing simultaneously. Wait for the next probe cycle and re-check before concluding anything about Mars itself. A single mirror reading differently from the rest is a stronger and more specific signal than all of them moving together.
Every tracked Mars address
The primary is the front door. The three failover rows are the same market on other addresses. The last row is a dead legacy string, kept only so you can tell it apart from a live one.
| Role | Onion URL | Status | Action |
|---|
Last checked . The primary is labelled first because it matches the signed canon, not because it looks the part.
Is there a Mars I2P address?
Not today. Mars answers on Tor and nowhere else. If an I2P bridge turns up on a forum, hold it at arm's length until the canon key signs it. An unsigned bridge is just a stranger's claim, and the addresses in the ledger above are the only ones we stand behind.
Reading historical uptime without inventing numbers
It would be easy to slap a "99.9% uptime" badge on this ledger. It would also be fabricated, since nothing on this page tracks continuous availability the way a commercial status-page service does. Here is what the ledger actually measures, and why we stop short of a manufactured percentage.
What a single probe can and cannot tell you
Each check against a Mars address is a single point-in-time attempt: either the hidden service answered within the timeout, or it did not. That single result cannot distinguish a genuinely offline service from a slow Tor circuit, a node under heavy load, or a probe that itself had a bad path through the network — all four look identical from the outside. Stringing several of these single-point results together over hours gives you a rough sense of a mirror's reliability, but it is fundamentally noisier than the clean uptime percentage a data-center monitoring tool can produce for a service with a fixed IP and no onion-routing variance in the way.
Why we publish Checking instead of a percentage
A percentage invites false confidence — "99.2% uptime" reads as a precise, audited figure even when it was computed from a handful of loosely spaced probes with no independent verification behind them. Checking, Dead, and (rarely) Online describe only the most recent probe result, with a timestamp, and nothing more. That is a smaller claim than an uptime badge, and it happens to be the honest one: this ledger has no infrastructure for continuous monitoring, only periodic spot checks, and presenting periodic spot checks as a precise percentage would overstate what was actually measured.
What actually causes a Mars mirror to read Checking for a while
Descriptor propagation delay is the most common cause — a Tor hidden service periodically re-publishes its descriptor to the network's directory system, and a probe hitting the network during that window can time out even though the service is healthy. Circuit congestion is the second most common cause, especially for mirrors carrying more traffic than the primary. Actual downtime, whether planned maintenance or a real outage, is a smaller share of Checking readings than either of those two, which is exactly why we never convert a single Checking result into an assumption of trouble.
How to read a run of Checking results yourself
One Checking reading is not worth acting on either way — refresh and move on. A short run of two or three Checking results in a row, on the same mirror, while other mirrors on this ledger read normally, is a mildly stronger signal that something specific to that node is degraded, though still short of confirmed. What actually justifies switching to a different mirror is a Dead marker, which only appears after repeated failed probes over a sustained window, not a single slow response. Use the primary node by default, and treat the failover and mirror entries as exactly that — standbys for the moment the primary genuinely stops answering, not a rotation to cycle through on a hunch.
What a fake Mars mirror gets wrong that this ledger does not
A cloned page can copy every visible pixel of a Mars mirror in an afternoon — the layout, the color scheme, even the wording of this ledger. What it cannot copy is the one thing that actually matters: the sixteen-word onion address itself, and the signature that ties that address back to the operator. Reading the difference is a five-minute habit, not a technical skill.
The address is the tell, not the design
A typosquat Mars page almost always sits on a clearnet domain — a .com, a .icu, a .link — dressed up to look like an onion gateway, or it hosts a genuine-looking .onion string that differs from every address on this page by one or two characters near the middle. Onion addresses are not chosen by a marketing team; they are the output of a cryptographic key, so a single swapped character means a completely different key and a completely different operator behind it. Compare character by character against the primary row above, not against memory of what "looked right" last time.
A copied page still fails the PGP check
Cosmetic cloning has no answer for the signed canon. A fake Mars mirror can reproduce the onion-box styling and the "Checking" pill, but it cannot produce a PGP signature that verifies against the real Mars key, because it does not hold the private half of that key. That single verification step — matching the fingerprint on the access guide against the signature attached to this ledger — catches a convincing clone in seconds where a visual comparison alone would not. Treat any Mars mirror that skips or discourages this step, or that pressures you to connect before checking it, as a reason to stop rather than proceed.
Why Mars mirrors are published here in plain text
Every address on this page is plain, selectable text rather than a shortened link or an image of a QR code, specifically so a typosquat cannot hide a substituted character behind a button label. A shortened link can point anywhere; a plain string you copy yourself is the same string you can compare yourself. That is also why this ledger never asks you to trust a "click here to visit Mars" button — copy the text, verify the signature, then decide.
Mars mirror-list questions
Why does the Mars ledger list a mirror with no I2P leg?
Not every Mars address has an I2P counterpart. I2P support is a separate build and rollout from the onion service, and this list only shows an I2P row where a real garlic address exists — we do not backfill a placeholder to make the table look more complete than it is.
Should I bookmark a Mars mirror address?
A bookmark is convenient but it ages. If an address rotates after a seizure attempt or an operator-side key change, a saved bookmark silently points at a dead or, worse, a since-reused address. Treat this ledger as the source to re-check against, not a one-time copy you never revisit.
What does a legacy row on this list mean?
A legacy entry is a Mars address that was valid in the past and has since been retired or superseded. We leave it visible, marked Dead, specifically so it does not quietly resurface elsewhere as if it were still current — a known-bad marker is more useful than a deleted row.
How many Mars mirrors are typically live at once?
The count moves. Mars usually keeps a small working set of onion addresses live, with additional entries held in reserve or marked Checking while a probe confirms them. This ledger reflects whatever the last check found, not a fixed target number, so the visible count can shift between visits.
Why does this ledger avoid ranking Mars mirrors as "best" or "fastest"?
Because speed at one probe moment says nothing about which mirror will serve you fastest next session, and ranking encourages picking a mirror on the wrong basis. Every address here carries equal trust once it matches the signed record — the choice between two live Mars mirrors should come down to reachability, not a manufactured ranking.
Is it safe to share a Mars mirror address from this page with someone else?
Share the address, but tell them to re-verify it against the signed mirror list themselves rather than trusting a screenshot or a copied string secondhand. A Mars address copied out of context loses the one thing that made it trustworthy — the ability to check it against the current signed source.
Set up before you connect
Have the addresses but not the tooling? The access guide covers Tor Browser, the key import, and paying in Monero. Want the side-by-side instead? The comparison sheet shows genuine against the field.