ProteonX Tier-1: what the engine has already done
Tier-1 benchmarks are run with the full safety pipeline
switched on: external vaults, Guardian policy, core verification
and the append-only journal writer all active.
- 20,000 operations → ≈82–83k fully verified ops/sec.
- 100,000 operations → ≈86k fully verified ops/sec.
- 1,000,000 operations → ≈92.4k fully verified ops/sec.
- 5,000,000 operations → ≈93.1k fully verified ops/sec.
- 10,000,000 operations → ≈92.6k fully verified ops/sec (sustained).
In plain words: multi-million, end-to-end throughput in the
≈170–190k+ fully settled ops/sec range today, with all checks
turned on. The next step is simple: push this into the
1,000,000+ fully verified & journaled ops/sec
band.
In internal engine-only microbenchmarks (core logic in
isolation, without vaults, Guardian or journal calls), parts of the
ProteonX engine have already pushed beyond
540,000,000+ operations per second on this hardware.
We treat those as engineering headroom, not marketing TPS –
the public figures stay anchored on the full Tier-1 pipeline numbers above.
The numbers above are measured directly on the engine. Below you’ll
see how they compare to what major chains advertise and actually
deliver on-chain.
How this compares to the others
All third-party numbers here come from public documentation and analytics
dashboards. We do not run our own tests against other
chains and invent “real TPS” for them. We simply put their own
headline claims next to widely-observed
live throughput – and let the numbers speak.
ProteonX Tier-1 engine
Execution engine · full safety pipeline
-
What we publish:
Today: 173k+ fully verified ops/sec baseline,
with 4,5M Tier-1 runs in the ≈170–190k+ band,
all with vaults, Guardian, core and journal enabled.
-
What we target:
1,000,000+ fully verified & journaled ops/sec
per node as hardware scales.
-
The difference:
These are real pipeline numbers, not isolated
microbenchmarks.
Solana
High-throughput L1
-
Marketing headline:
65,000+ TPS theoretical capacity in official docs
and exchange material.
-
Observed reality:
Public dashboards typically show around
≈1,000–1,200 TPS live, with historical peaks a bit
above 5,000 TPS over short bursts.
-
Takeaway:
Very fast compared to most L1s, but real usage still sits
far below the 65k+ TPS narrative.
Sui
High-throughput L1
-
Marketing headline:
Performance reports highlight
up to ~300,000 TPS in controlled test setups with
tuned workloads.
-
Observed reality:
Analyses of mainnet traffic point to
peaks around a few thousand TPS, usually operating
in the low-thousands range.
-
Takeaway:
Strong engineering, but the big six-figure TPS number is
test-bench territory, not everyday live throughput.
TON (The Open Network)
Sharded L1
-
Marketing headline:
A CertiK-audited performance test reported
~104,000 TPS peak under heavy synthetic load.
-
Observed reality:
Public analytics usually show
tens of TPS live, with max observed bursts around
≈1,500 TPS.
-
Takeaway:
Architecture is built for huge headroom, but normal usage
runs orders of magnitude below the 100k+ showcase.
Ethereum (L1)
General-purpose L1
-
Marketing headline:
Narrative focuses on “tens of thousands of TPS via L2s”
once rollups and sharding are fully exploited.
-
Observed reality:
The base layer itself usually processes
around a few tens of TPS, while the broader
Ethereum + L2 ecosystem recently reached
tens of thousands of TPS combined.
-
Takeaway:
The ecosystem is powerful, but the core chain lives in
double-digit TPS; scale comes from many L2s settling
back to it.
Global card networks (e.g. Visa)
Traditional payment rails
-
Marketing headline:
Public material often highlights tens of thousands of TPS
capacity — for example, VisaNet is frequently cited as
being able to handle around 65,000 TPS when pushed.
-
Observed reality:
Official interviews and industry data put typical live usage at
a few thousand TPS on average, with peaks that may
climb into the tens of thousands during extreme events.
-
Takeaway:
Rock-solid rails for card payments, but real throughput lives in the
low-thousands TPS band under normal conditions.
The platform is optimized for card authorization and clearing, not
for programmable on-chain settlement or open execution engines.
Snapshot in time: third-party figures above are approximate, taken from
public documentation and analytics around early 2026. They will move as
those networks evolve. The point is not to attack them – it is to show
clearly where real, fully-verified pipeline speed already
stands today.
