Skip to content
Crypto Almanac Daily
G
Perp Dex

GMX V2 Perps

Best for audited, multi-chain deployment

Total value locked:$175m at this checkChains:4Audits recorded:2Audit report linked:YesRubric:v2.0 · verified 9 Aug 2026
Kayla PetersonKayla PetersonDeFi Research Analyst· Last verified August 9, 2026
Confidence ARubric v2.0Verified August 9, 2026
9.3
out of 10
Open account
Scorecard

How it rates

Counterparty & contract risk · 30%10.0
Cost transparency · 20%9.0
Market quality · 25%8.0
Transparency & track record · 15%10.0
Public documentation surface · 10%10.0
Pros
  • Audit report linked from a public dataset
  • Documentation published at a public address
  • Deployed across 4 chains
Cons
  • Several indicators could not be verified from public sources at this check
How this score was built

Each indicator scores 2, 1 or 0. A pillar is the points earned over the points available; the overall score is the weighted sum. Every source below is public — check any of them yourself.

Counterparty & contract risk · 30% weight10/10 points · 10.0/10
Cost transparency · 20% weight9/10 points · 9.0/10
  • Met
    Documentation published where fees are described

    Documentation reachable for fee review.

    docs.gmx.io/
  • Met
    Application reachable without an account

    No account required to reach the app.

    gmx.io
  • Met
    Independent dataset publishes protocol economics

    TVL and change history published independently.

    defillama.com/protocol/gmx-v2-perps
  • Partial
    Fee model documented publicly

    Documentation available; specific fee tiers not captured at this check.

    docs.gmx.io/
  • Met
    No account required to read protocol documentation

    Open documentation.

    docs.gmx.io/
Market quality · 25% weight8/10 points · 8.0/10
Transparency & track record · 15% weight10/10 points · 10.0/10
Public documentation surface · 10% weight10/10 points · 10.0/10

An oracle-priced perpetuals protocol using isolated per-market liquidity pools, holding $175m across four chains with published audit reports.

Our assessment

GMX v2 is the highest-scoring perpetuals venue in this comparison: $175m across four chains, two recorded audits with reports linked, and a design that isolates each market's liquidity rather than pooling all risk together.

Oracle pricing instead of an order book

Trades execute at an oracle price rather than against resting orders, so there is no slippage from book depth and no need for market makers to quote. The trade-off is a direct dependency on the oracle: its latency, its sources and its failure modes become the protocol's failure modes, which is why oracle manipulation has been the recurring attack on designs of this kind.

Isolated GM pools contain the damage

In v1, one shared pool backed every market, so a loss on any one asset hit every liquidity provider. v2 gives each market its own pool. A provider chooses which exposures to underwrite, and a blow-up in a thin market does not reach the pools backing major ones. That is a direct response to how the first design actually failed.

Liquidity providers are the counterparty

Depositors take the other side of every trade in their pool. When traders are net profitable, providers lose — the yield is a payment for underwriting directional risk, not a fee for supplying capital. Anyone depositing should understand they are running a short-volatility book.

Who it suits

GMX fits traders wanting deep on-chain perps without an order book, and providers who understand they are the house. Traders who want order-book execution and price-time priority should compare dYdX or Hyperliquid.

Reference

Frequently asked

Does this score mean GMX V2 Perps is safe to use?

No. It measures what an outsider can verify: linked audits, published documentation and independently recorded market data. Contract behaviour is not tested by us.

Where does the TVL figure come from?

An independent public analytics platform, queried at the verification date. Anyone can re-run the same query.

Why do some protocols score zero on audits?

Because no audit report is linked in the public dataset. It records what a user can reach, not a claim that no audit was ever performed.