OBEX v2

Deterministic Infrastructure for Onchain Finance 0x2dB219A4C16f1bae611361d21421fDe06f447f47

[OBEX v2 @ObexProtocol]

OBEX IS BACK.

Eight months ago, I started rebuilding.

Today, OBEX launches a new chapter on Robinhood Chain.

OBEX is deterministic infrastructure for onchain finance. In simple terms, it moves critical decisions out of the hands of whoever happens to control the next block and puts them into explicit protocol rules.

Participation is determined through RAM-hard Proof-of-Effort and verifiable randomness. Blocks are validated through deterministic rules. Transactions follow canonical ordering, removing discretionary reordering as a source of MEV. Rewards are calculated independently of who proposes the block.

Same inputs. Same rules. Same result.

But why did OBEX disappear?

The original project launched on Solana. It reached a $1.7M ATH, but I didn’t have the resources to keep funding the infrastructure, development and operating costs required to take it further.

So the project was abandoned.

The idea wasn’t.

Over the last eight months, OBEX has been rebuilt from the ground up and brought to production readiness.
I chose Robinhood Chain as the new home because its focus on programmable financial infrastructure and tokenized assets makes it a natural environment for what OBEX is designed to become.

Today, I am launching a new native $OBEX token on Robinhood Chain.

This is not a migration of the old token. It is a completely fresh deployment and a new beginning for the protocol.

The first OBEX proved the idea could attract attention.

This one is built to actually go somewhere.

One final note for the community:

Unfortunately, I was locked out of the original OBEX X account and no longer have access to the email connected to it, as it has been quite some time since the original project was active.

To verify that this is the same person behind the original OBEX, I’ll be using the original OG Telegram group/channel as proof of continuity.

The old account may be gone.

I’m not. OBEX isn’t.

OBEX is live.
CA: 0x2db219a4c16f1bae611361d21421fde06f447f47

[OBEX v2 @ObexProtocol]

Five days left in epoch 1.

The board right now: 8 wallets, 73 verified proofs, 7 wallets already maxed at all 10 slots.

That last number is the one worth sitting with. Almost everyone who starts mining OBEX runs it to the cap. They do the work, watch it verify, and immediately take every slot they can.

612 million BLAKE3 labels stand behind that record. Every proof forced 256 MiB of real RAM and passed 16 cryptographic challenges before it counted. Nothing self reported, nothing you have to take our word for: https://t.co/zkHfmaJk6H

Sep 1 the record freezes. The pool splits pro rata across every eligible proof, and whatever is on the board at that moment is what gets divided. The wallets already maxed have locked their claim. The only open question is how many ways it splits.

500k $OBEX and about ten minutes gets you all 10.

https://t.co/nwVemmZ9b1

[OBEX v2 @ObexProtocol]

People ask why OBEX came back on Robinhood Chain instead of where it started. Simple: the design finally matches the rails.

OBEX mining is real RAM-hard work done on your own machine. The chain's only job is to be the truth layer: who holds the ticket, who locked what, who owns which proof. EVM is the best truth layer in crypto and it isn't close.

$OBEX is a standard ERC-20. That one fact does a lot of work. The holder gate is a single balanceOf call anyone can reproduce. The dev lock until 2069 sits in a standard vesting contract you can read on Blockscout. Every wallet supports it, every explorer indexes it, every DEX can pool it. Nothing custom, nothing you have to trust us on.

And Robinhood Chain isn't a random EVM. It's an Ethereum L2 built for tokenized equities and ETFs, real assets living as ERC-20s with Chainlink prices on chain. OBEX is deterministic protocol infrastructure. Deterministic rules plus real financial assets on the same rails is exactly the pairing we designed for.

The old chain got a token. This chain got a protocol.

https://t.co/VkRBnsIW6O

[OBEX v2 @ObexProtocol]

Quick read of the board this morning.

8 wallets mining. 7 of them maxed out all 10 slots.

Think about what that means. Almost nobody who starts mining OBEX stops before the cap. They do the work, see it verify, and immediately run it to the limit.

73 proofs on the record. Over 600 million BLAKE3 labels computed and verified. Every single one public and checkable forever: https://t.co/zkHfmaJk6H

The window closes Sep 1 and the pool splits pro rata. The wallets already maxed have locked their share. The only question left is how many ways the pool splits.

500k $OBEX and about ten minutes gets you the full 10 slots: https://t.co/nwVemmZ9b1

[OBEX v2 @ObexProtocol]

Your proof is fine, the copy step just failed. The clipboard only had 6 characters in it when you hit paste, so PowerShell didn't actually load the file.

Fix: in PowerShell, cd into the folder where you ran the miner (obex-proof.bin lives next to obex-miner.exe), then run the Step 3 command from the dashboard again. If it prints a red error, you're in the wrong folder.

Quick check that it worked, run this after:
(Get-Clipboard).Length
It should say around 50,000, not 6.

Then hit Paste from clipboard on the dashboard and submit.

[OBEX v2 @ObexProtocol]

Some math on why OBEX proofs can't be faked.

Each proof forces 8,388,608 BLAKE3 labels. Exactly 256 MiB of memory, label by label, no shortcuts.

Then the verifier hits you with 16 random challenges. It picks 16 labels out of the 8.4 million and demands you open them.

Skip half the work? Each challenge catches you with a coin flip. Passing all 16 is 1 in 65,536.

Skip even 10% of the labels? You still get caught 81% of the time.

The only winning strategy is doing all the work. That's the whole design.

https://t.co/Hhs3oy2pUP

[OBEX v2 @ObexProtocol]

The fairness math on epoch 1, live numbers:

The cap is 10 proofs per wallet. A whale with 100x your tokens gets the same 10 slots you do.

The pool splits pro rata: your proofs divided by all proofs on the record.

Right now the record holds 73 proofs, so a maxed wallet owns 10/73 of the pool. That's 13.7% of epoch 1, and it dilutes as more miners join.

Every day you wait, the denominator grows.

612 million BLAKE3 labels already stand behind that board. The window closes Sep 1.

https://t.co/nwVemmZ9b1

[OBEX v2 @ObexProtocol]

How OBEX picks who makes each block, with zero communication between nodes:

Every miner's proof contains beta, a 64-byte VRF output. Unbiasable. Fixed before any block exists.

For height h, every node computes score = H(beta, h) for every member of the record and takes the lowest. That's the leader. Done.

The chance of a tie between two scores is 1 in 2^256. The universe ends first, and there's a tiebreaker anyway.

Same record, same math, same leader, on every node independently. Nothing to vote on. Nothing to fork.

Consensus isn't reached. It's computed.

https://t.co/Hhs3oy2pUP

[OBEX v2 @ObexProtocol]

The most underrated number in OBEX:

Mining a proof takes about a minute of RAM-hard compute.

Verifying it takes milliseconds.

Each of the 16 challenges opens a Merkle path 23 hashes deep (2^23 = 8,388,608 labels). That's 368 hashes plus one VRF check, and the proof is either perfect or garbage.

Which means anyone, on any laptop, can re-verify every proof on the record from raw bytes. Today, next year, forever. No trust, no archive node, no "just believe us."

Miners do the work once. The world can check it for free, forever.

https://t.co/zkHfmaJk6H

[OBEX v2 @ObexProtocol]

The record keeps growing.

73 verified mines. 8 wallets on the board. Over 600 million BLAKE3 labels computed, verified, and public.

Every entry is 256 MiB of forced RAM work, checked cryptographically at submission. Nothing self reported. The board doesn't care who you are, only what you computed: https://t.co/zkHfmaJk6H

Window closes Sep 1. The pool splits pro rata, so every slot claimed now is a bigger share of the record.

500k $OBEX gets you in: https://t.co/nwVemmZ9b1

[OBEX v2 @ObexProtocol]

@krabat587 Currently dashboard does not support GMGN wallets directly. You would need to import your GMGN wallet to Metamask

[OBEX v2 @ObexProtocol]

A lot of you have been asking: can you mine OBEX from your phone?
Yes. 📱

The miner itself needs a real machine because every proof forces 256 MiB of RAM. That memory requirement is what makes the work meaningful.
Your phone can simply drive a free Linux environment in the cloud.

Open https://t.co/VhIQ0DuvGa on your phone, log in with any Google account and run:
curl -fsSL https://t.co/gYSNdQC7wQ | sh

Then head to https://t.co/xa9xPERogl, connect your wallet and meet the 500K $OBEX minimum. The dashboard gives you the exact mining command.

Paste it into the terminal, wait about a minute and your proof will be generated.

Copy the proof back into the dashboard, submit it and the protocol verifies it instantly.
Verified → mined → on the record.

The dashboard also has a batch command that runs all 10 mining slots in one go.

No mining rig required.

Your phone can be the remote control.

[OBEX v2 @ObexProtocol]

Epoch 1 closes September 1. Here’s what comes after.

When Epoch 1 closes, the record freezes.

Every eligible proof becomes permanent, the pool is distributed pro rata, and that frozen record becomes the foundation for what comes next.

Epoch 2 is different.

Getting in will cost more than simply holding $OBEX.

That’s all I’ll say for now.

But if you understand what that means for supply, you already understand.

The engine that turns the record into a chain is already written. The spec and working Rust implementation are public at https://t.co/Hhs3oy2pUP⁠.

We don’t promise infrastructure.

We publish it.

And the thing it stands on is being mined right now.

CoinGecko application is currently in review.

And understand what Epoch 1 actually is.

The miners on that board aren’t just early users.

They’re part of the genesis data.

Mine OBEX: https://t.co/mCapT3mjqi

[OBEX v2 @ObexProtocol]

CoinGecko application submitted.

Request ID CL2508260006, pending review.

Verified contract, locked dev supply til 2069, live liquidity, working miner, public records.

Epoch 1 closes Sep 1. Pro-rata pool.

500k $OBEX gets you in: https://t.co/YEyVJDzQSw

[OBEX v2 @ObexProtocol]

Every chain that launches mining dies in the same spot.

Node consensus.

Nodes have to agree on who makes the next block. Votes, timing rounds, negotiation between machines. That one step has killed more chains than anything else in crypto.

OBEX deleted the step.

There is no vote in v2. The mining record being built right now IS the consensus. When the window freezes, who produces each block is pure math on that record. Every node runs the same math alone and gets the same answer. Two valid blocks at the same height cannot exist. Nothing to fork, nothing to argue about.

And this isn't a pitch, it's published. The spec and the working Rust implementation are live at https://t.co/Hhs3oy2pUP. Read how a block gets validated before a single node is online.

63 proofs already verified on the record. It freezes Sep 1. Everything after stands on it.

We didn't solve the hardest problem in mining chains.

We removed it.

https://t.co/VkRBnsIW6O

[OBEX v2 @ObexProtocol]

$OBEX UPDATE: Working miner you can download today. Windows and Linux, straight from https://t.co/VkRBnsIW6O

63 proofs mined and cryptographically verified. 7 wallets on the record, 6 already maxed out their 10 slots

Over half a billion BLAKE3 labels computed. 256 MiB of RAM forced per proof. Nothing self reported, every entry re-verifies from its bytes forever

Consensus design published. No votes between nodes, block production is pure math on the mining record. Spec and Rust code live at https://t.co/Hhs3oy2pUP

Dev supply locked on chain until 2069. Non cancelable

Epoch 1 closes Sep 1. The pool splits pro rata over eligible proofs, so every slot claimed now is a bigger share of the record

All of it checkable right now: https://t.co/VkRBnsIW6O

[OBEX v2 @ObexProtocol]

@cobramotion Done ✔️

[OBEX v2 @ObexProtocol]

MINING UPDATE

Epoch 1 is filling up.

63 verified mines on the record
7 wallets ranked
6 already maxed at 10/10

Every entry is real RAM-hard work. 8,388,608 BLAKE3 labels per proof, 256 MiB forced, checked cryptographically the moment it hits the dashboard. Over half a billion labels computed so far. Nothing on the board is self-reported.

Allocation is pro-rata. Your eligible proofs divided by everyone's. The more of the record you hold, the more of the pool you take.

Window closes Sep 1, 00:00 UTC. 6 days.

Hold 500k $OBEX, run the miner, claim your slots:
https://t.co/nwVemmZ9b1

The board is public. Watch it live:
https://t.co/zkHfmaJk6H

[OBEX v2 @ObexProtocol]

The ticket. The work. The record.

Native OBEX isn’t bought.

It is mined.

The live campaign at https://t.co/VkRBnsIW6O has three parts.

THE TICKET

Hold at least 500,000 $OBEX on Robinhood Chain.

Checked on-chain when you connect and again at submission.

No ticket. No mining.

Running multiple wallets means acquiring multiple tickets. Sybil pressure is priced into the protocol rather than policed by IP addresses.

THE WORK

Install the miner:

curl -fsSL https://t.co/gYSNdQC7wQ | sh

Create your identity once.

Mine for roughly 10 seconds.

Submit your proof at the dashboard.

THE VERIFICATION

Every proof is structurally validated and cryptographically verified.

ECVRF.
Merkle openings.
Label recomputation.

Valid proofs become:

ELIGIBLE ✓

Invalid proofs are rejected with the reason.

Current floors:

256 MiB minimum dataset.
16 minimum challenges.
10 proofs maximum per wallet this epoch.

THE RECORD

Every verified proof is recorded against your wallet.

That record qualifies holders for the launch allocation of native OBEX when the network activates.

The $OBEX you hold today is the ticket in.

It is not the payout.

Every entry is independently verifiable from its proof bytes, making the allocation record auditable.

Mining window closes September 1, 00:00 UTC.

The math does not negotiate.

https://t.co/mCapT3mjqi

[OBEX v2 @ObexProtocol]

10 seconds to prove. 438μs to verify.

From the reference implementation:

256 MiB dataset
16 challenges

Mining: 8–10 seconds
Verification: 438 microseconds
Proof size: 37,428 bytes

That’s roughly a 20,000:1 asymmetry.

Why is verification so cheap?

The dataset contains 8,388,608 labels.

The verifier doesn’t need to rebuild it.

It checks the 16 challenged openings, their Merkle paths, the label equations and the ECVRF proof.

About ~1,100 hashes.
16 label-equation recomputations.
1 ECVRF verification.

That’s it.

The verifier holds only the ~37 KB proof.

And verification is total.

Change a single byte and something breaks:

Merkle path.
Label equation.
VRF proof.
Root commitment.

The proof either satisfies the protocol or it doesn’t.

You don’t have to trust our server either.

OBEX ships obex-verify, a headless verifier.

Feed it the proof bytes.

Get the verdict.

Same bytes → same answer.

There is even a frozen golden vector for independent implementations to reproduce byte-for-byte.

Determinism isn’t a slogan. It’s a test that fails.

[OBEX v2 @ObexProtocol]

Commit first. Draw second. No free rerolls.

After building the dataset, you commit to every label with a Merkle tree.

One 32-byte root represents the entire dataset.

Then OBEX derives your draw:

alpha = H_vrf(chain_tag, LE64(epoch), pk, root)

beta = ECVRF(sk, alpha)

Using ECVRF-RISTRETTO255-SHA512 (RFC 9381).

The output is deterministic for your key and verifiable by anyone.

Now the important part:

Your VRF input includes the dataset root.

Want a different draw?

You need a different alpha.

Different alpha → different root.

Different root → rebuild the entire dataset.

There is no free reroll.

The resulting beta determines which labels you must reveal.

You don’t know the challenges until after you’ve committed to the dataset.

For every challenge, you provide the required labels and Merkle paths. The verifier independently reconstructs the equations and checks them against the committed root.

Then eligibility comes down to one deterministic comparison:

beta[0..32] < threshold

No operator decides.

No committee decides.

The math decides.