Skip to main content
CoinPulse
Perpetual DEXsNot published

GMX review: why it scores 7.04

GMX scores 7.04 out of 10 and ranks #1 of 14 in the perpetual DEXs category, with 100% of the category's criteria measured as of the October 19, 2026 collection. Its standout result is the combination of scale and audits: $213.0M in value locked, about seven times the category median, alongside one published audit link on DefiLlama where the median perpetual DEX has none. The points it misses are spread out: the single audit link scores 6 and costs 1.04, the $213.0M figure lands in the 7 band on a log scale and leaves 0.90 on the table, and no public status page was found, which scores 2 and costs 0.64.

Last checked Analysis written 0 affiliate links About GMX
7.04
out of 10
100% of the score measured
#1 of 14
Where the 7.04 points come from
Scale of value trusted 2.1Track record 1.6Security audits 1.6Multi-chain reach 0.8Latency 0.8Status transparency 0.2

At a glance against other perpetual dexs

Ahead of the category
  • Security audits 6/10 median 2.0 (+4.0)
    1 published audit with links (DefiLlama) — read the findings yourself.
  • Scale of value trusted 7/10 median 6.0 (+1.0)
    $213.0M in value locked (DefiLlama).
  • Track record 9/10 median 8.5 (+0.5)
    Live 3.2 yr (time since DefiLlama listing (a floor on true age)).
Behind the category

No measured criterion is clearly below the median.

Why CoinPulse rated GMX 7.04 in perpetual DEXs

The perpetual DEXs ranking weights six criteria: Scale of value trusted (30%), Security audits (26%), Track record (18%), Multi-chain reach (10%), Latency (8%) and Status transparency (8%). A perpetuals venue needs liquidity behind the book and an audited engine, so value locked and audits carry more than half of the weight between them. Security record is not scored in this category, so it counts neither for nor against GMX.

All six criteria were measured, so coverage is 100%. The contributions: Scale of value trusted 7 × 30% = 2.10, Security audits 6 × 26% = 1.56, Track record 9 × 18% = 1.62, Multi-chain reach 8 × 10% = 0.80, Latency 10 × 8% = 0.80, Status transparency 2 × 8% = 0.16. Total: 7.04. The 2.96 points it did not get break down as 1.04 on audits, 0.90 on scale, 0.64 on status transparency, 0.20 on reach and 0.18 on track record. The methodology page sets out the bands behind each score.

Where GMX scored well

Scale of value trusted: $213.0M locked (7/10)

DefiLlama recorded $212,994,372 in total value locked for GMX at collection, against a category median of $30.5M. The criterion uses a log scale, so that lead places GMX one band above the category median score of 6 rather than far ahead of it. Value locked shows how much capital users keep inside the protocol's contracts; it is not loss protection. A pool of this size can still be drained by a contract bug or an oracle failure, and nothing in the figure insures a position.

Security audits: one published audit link (6/10)

DefiLlama lists one published audit link for GMX, pointing to the audits folder of its GMX Synthetics repository, which scores 6. The criterion counts the links DefiLlama lists, so a single link to an audits folder does not mean GMX has had only one audit. It is still the widest gap between GMX and its category: the median perpetual DEX in this run has no listed audit and a median audit score of 2. At 6, though, audits are also where GMX gives up the most points (1.04). The criterion counts audits that can be read, not their findings or scope, so the number says nothing on its own about what the auditors found or whether code shipped since was reviewed. Audits reduce risk; they do not remove it.

Track record, reach and latency (9, 8 and 10 out of 10)

GMX has been live 3.2 years by DefiLlama's listing date (1,167 days, against a category median of 1,080.5). The methodology treats a listing date as a floor on true age, so the protocol may be older than this figure. DefiLlama also lists it on four chains, against a median of 2.5; reach carries only 10% of the weight and works as a tie-breaker.

Our probe timed the gmx.io interface at a median 146 ms to first byte, which scores 10 inside this category. That is one location, with the slowest of three samples discarded, and the category median is 131.5 ms, so GMX is fast but not unusually so among its peers. The probe reads the web front end, not the contracts, which keep running whether or not the website answers.

Where GMX lost points

No public status page (2/10)

The collection found no public, machine-readable status page for GMX at its documented or conventional hosts. That scores 2 and is the third-largest deduction, after audits and scale, worth 0.64 of the 0.80 points available. The methodology treats this as a fact about the service, not an excuse: a venue you can audit from outside beats one you cannot, even if the opaque one happens to be reliable. It is also where GMX trails the runner-up, Aevo, which has a status page without a parseable feed and scores 5. Two caveats apply. The site reads three status platforms, so a page on a fourth would be missed, and the category median for this criterion is also 2, so GMX is typical here rather than unusual.

Scale stops at the 7 band (0.90 points)

Because the scale criterion carries 30% of the weight, the three bands between a 7 and a 10 cost GMX 0.90, second only to audits. $213.0M is well above the category median, but on a log scale the next bands belong to venues holding billions; Hyperliquid, with $6.78B locked, scores 9. This is a measured difference in capital held, not a judgment on how the protocol operates.

What the GMX rating does not tell you

Trading fees, funding rates, price impact, available markets, position limits and the interface experience are not scored for any perpetual DEX on CoinPulse. The method also does not read liquidity depth per market or how the venue behaved during volatile sessions. Smart-contract risk is only approximated through the audit count and the value locked, and oracle risk, which matters more for a perpetuals venue than for a spot DEX, is not measured at all. For what GMX is and how it works, see the GMX company profile.

For context only: DefiLlama's hack database showed no incident matched to GMX when checked on October 19, 2026. Security record is not part of the perpetual DEX profile, so that finding did not add to the score, and incidents outside that database are not scored either. The absence of a status feed also means the site logged zero incidents for GMX over the 90-day window, which reflects a lack of data rather than a measured run of clean uptime.

How GMX compares in the perpetual DEXs ranking

All 14 ranked perpetual DEXs have 100% coverage, so the gaps come straight from criterion scores. Aevo, second at 6.68, matches GMX on audits, track record, reach and latency. The 0.36-point difference is two criteria pulling in opposite directions: GMX gains 0.60 on scale (7 against 5, with $14.7M locked at Aevo) and gives back 0.24 on status transparency (2 against 5).

Hyperliquid, third at 6.60, is the clearest case of the weights at work. It holds $6.78B, scores 9 on scale and gains 0.60 points there, but DefiLlama lists no published audit for it, so it scores 2 against GMX's 6 and loses 1.04 on the heaviest remaining criterion, leaving it 0.44 behind. Gains Network, fourth at 6.24, has the same audit score as GMX and one more chain, but $8.7M locked scores 4 on scale, which costs it 0.90, for a net gap of 0.80. The gap below the top four is modest: Jupiter Perps and MUX Protocol tie at 6.16 in fifth and sixth, and seventh-placed dYdX scores 5.84 with no listed audit.

What the 7.04 means in practice

GMX leads this category because it is the only venue in the top five that scores well on both scale and audits at once; every peer close to it is strong on one and weak on the other. The points it misses come from an audit score of 6 built on a single listed link, a status criterion where almost the whole category scores 2, and a value-locked figure that is high for a perpetual DEX but far from the billions that earn the top bands. The rating does not tell you whether a particular market, position or oracle feed is safe, and it says nothing about cost of trading. Reading the linked audit reports and checking the contracts you interact with is still your job. No service can pay for its position here; see how this site is funded.

Criteria vs the category

The bar is GMX’s mark; the white tick is the median across the services we rank in the category. Open a criterion for the reasoning and the evidence.

Scale of value trusted
30% of the score
+1.0 vs median7/10
Category median 6.0Why this score ↓

$213.0M in value locked (DefiLlama). Capital is the market voting with its money — scored on a log scale, so the jump from $10M to $100M counts for more than $10B to $20B.

Measured 2026-10-19 14:40 UTCCheck the evidence ↗
Security audits
26% of the score
+4.0 vs median6/10
Category median 2.0Why this score ↓

1 published audit with links (DefiLlama) — read the findings yourself. Audits reduce risk, they do not remove it.

Measured 2026-10-19 14:40 UTCCheck the evidence ↗
Track record
18% of the score
+0.5 vs median9/10
Category median 8.5Why this score ↓

Live 3.2 yr (time since DefiLlama listing (a floor on true age)). Most failures happen young, so time survived is real signal.

Measured 2026-10-19 14:40 UTCCheck the evidence ↗
Multi-chain reach
10% of the score
8/10
Category median 8.0Why this score ↓

Deployed on 4 chains (DefiLlama). Breadth is a minor tie-breaker — a focused single-chain protocol can still be excellent.

Measured 2026-10-19 14:40 UTCCheck the evidence ↗
Latency
8% of the score
10/10
Category median 10.0Why this score ↓

Median 146 ms to first byte from the live interface, our own probe, slowest sample discarded, single location. Compared only within this category.

Measured 2026-10-19 14:39 UTCCheck the evidence ↗
Status transparency
8% of the score
2/10
Category median 2.0Why this score ↓

No public status page found. Nothing about its reliability can be verified from outside — including by us.

Measured 2026-10-19 14:39 UTC

What we measured

Value locked
$213.0M
Median $30.5M · DefiLlama
Published audits
1
Median 0 · DefiLlama
Track record
3.2 yr
Median 3.0 yr · public record
Chains
4
Median 3 · DefiLlama
Latency
146 ms
Median 132 ms · our probe

Ranked nearby in perpetual dexs

Full perpetual dexs ranking

About this rating

Is GMX safe?+

No rating can guarantee that, and this one measures operational evidence from public data, not safety. For GMX specifically, the October 19, 2026 collection found $213.0M locked, one published audit link on DefiLlama, 3.2 years since its DefiLlama listing and no incident matched to it in DefiLlama's hack database. Audits and value locked reduce risk without removing it, and oracle risk, market-level liquidity and front-end risks are not measured.

Why is GMX ranked #1 of 14 perpetual DEXs?+

It scores 7 on scale and 6 on audits, and no other venue in the category pairs both. Aevo matches it on audits but holds $14.7M, scoring 5 on scale; Hyperliquid holds far more at $6.78B but has no listed audit, scoring 2. Those two criteria carry 56% of the weight, so GMX finishes 0.36 ahead of Aevo and 0.44 ahead of Hyperliquid.

How many security audits does GMX have?+

DefiLlama lists one published audit link for GMX, which scores 6 out of 10. The link points to the audits folder of the GMX Synthetics repository, so it is not a claim that GMX has only one audit. The criterion counts listed links, not findings, so the reports themselves are worth reading before relying on the number.

Why did GMX score 2 on status transparency?+

The collection found no public status page with a machine-readable feed. That is scored, not excused, because a service that cannot be checked from outside is harder to trust. The site reads three status platforms, so a page hosted elsewhere would be missed; the category median is also 2.

Are GMX fees part of the rating?+

No. Fees, funding rates, slippage, position limits and the interface are not scored for any perpetual DEX on CoinPulse, because an honest reading of them needs repeated real trades that the method does not perform.

Evidence and updates

Written October 19, 2026 from the October 19, 2026 data collection. Scores are recalculated on every collection; the analysis is reviewed when the score or rank moves.