Cortex

About Cortex (CTXC)

What Is Cortex?

Cortex is a proof-of-work Layer 1 blockchain built to run machine learning models inside smart contracts. Its virtual machine, the Cortex Virtual Machine (CVM), is ported from the Ethereum Virtual Machine and adds instructions that call AI models stored on-chain, so an inference result is produced and verified by network consensus rather than supplied by an off-chain service. CTXC is the chain's native coin: it pays for transactions and model calls, rewards miners, and rewards the developers whose models get used.

The project matters as one of the earliest attempts at "on-chain AI" in the literal sense. Most AI and crypto projects today connect blockchains to models that run elsewhere; Cortex executes the model itself as part of block validation, which makes results reproducible and auditable but also limits how large and complex those models can be.

Who Founded Cortex? (History)

Cortex is developed by Cortex Labs; the project website carries a copyright line for NeoCortex Global Limited, British Virgin Islands. The project's published materials do not give a detailed team roster, so readers should treat founder information from third-party sources with care.

Key documented milestones:

• 2018 token sale. CTXC was first issued as an ERC-20 token on Ethereum. A single private placement ran from February 7 to March 7, 2018, at 1 ETH = 1,500 CTXC, with a cap of 40,000 ETH for 60 million CTXC (about 20% of the total). Bitmain and FBG Capital are named as lead investors.
• 2019 mainnet. Cortex launched its own chain in mid-2019. ERC-20 holders migrated by sending tokens to a burn address on Ethereum and receiving native CTXC, or through exchanges that handled the swap. A wallet provider's notice dated July 10, 2019 reported the mainnet launch complete.
• Layer 2 work. The current website highlights ZkMatrix, a zkRollup design for proof-of-work, EVM-compatible chains. Its documentation describes a testnet.

How Cortex Works

Cortex combines an Ethereum-style account and contract model with GPU-oriented extensions:

1. Cortex Virtual Machine (CVM) The CVM is backward compatible with the EVM, so standard Solidity contracts run on it. It adds infer instructions (Infer, opcode 0xc0, and InferArray, 0xc1) that let a contract call a stored model.

2. Synapse, the deterministic inference engine Different GPUs can produce slightly different floating-point results. Synapse (also called the CVM Executor) uses fixed-point execution and a deterministic operator library so every node gets exactly the same inference output.

3. MRT quantization MRT is a framework that converts MXNet models into fixed-point models the CVM can execute. Developers train normally, then quantize with MRT before uploading.

4. Storage layer Models and data are uploaded to the chain's storage layer, with a per-block upload quota (documented as 64k per block), a model size limit of 1 GB and a maturity period of 100 blocks before a model can be used.

5. Endorphin fees Endorphin is the unit that prices computation, similar to gas on Ethereum. Every transaction, contract call and inference consumes Endorphin, paid in CTXC.

Key Ecosystem Components

• CortexTheseus, the official Go client, and the tech documentation on GitHub.
• Cerebro, the block explorer at cerebro.cortexlabs.ai.
• Cortex Wallet, which the project lists for Chrome, iOS and Android; MetaMask can also connect through a custom RPC.
• Cortex Remix, a browser IDE adapted for AI contracts.
• ZkMatrix, the Layer 2 zkRollup project (docs).

CTXC Token Utility

CTXC is used to:

• Pay Endorphin fees for transfers, contract execution and AI inference.
• Reward miners, who receive a fixed block reward plus a share of fees.
• Reward model providers, who receive part of the Endorphin paid whenever their model is called. This is the incentive meant to attract AI developers to publish models openly.

Fixed tokenomics from the project's documentation:

• Total supply: 299,792,458 CTXC (the figure matches the speed of light in metres per second, most likely a deliberate reference).
• 150,000,000 CTXC reserved as mining rewards.
• 149,792,458 CTXC pre-allocated at genesis.
• Block reward: 7 CTXC, halving every 4 years.

What Makes Cortex Unique?

The distinguishing feature is that model inference is part of consensus. When a contract calls a model, every validating node runs it and must arrive at the same answer. That gives a trust property oracle-based AI integrations do not have: nobody has to trust a third party's claim about what a model returned. The cost is that models must be quantized, size-limited and run deterministically, which rules out today's very large models.

How Cortex Differs from Similar Chains

• Versus Ethereum and other EVM chains: Cortex runs Ethereum contracts but adds native model execution, which standard EVMs lack.
• Versus AI networks such as Bittensor or the Artificial Superintelligence Alliance: those coordinate or reward AI work that runs off-chain. Cortex puts the model and its execution on the ledger itself.
• Versus GPU compute marketplaces such as Render: Render sells GPU time for off-chain jobs; Cortex uses GPUs for on-chain execution and for mining.

Use Cases of Cortex

• Smart contracts that make decisions from a model, for example classification or scoring inside a DApp.
• Publishing open models that others can call and pay for per use.
• Verifiable inference where the reproducibility of a result matters more than model size.
• Experiments with zk-based scaling for proof-of-work EVM chains through ZkMatrix.

How Cortex Is Secured

Cortex uses proof of work based on Cuckoo Cycle, a memory-bound, graph-theory algorithm; the chain uses the CuckAroo 30 variant, designed to resist ASICs and favour GPUs. Difficulty targets a 15-second average block time. Like any proof-of-work chain, its security depends on total hashrate: an attacker controlling more than half of mining power could reorganise recent history. Because Cortex has its own mining algorithm rather than sharing one with a larger chain, its hashrate is independent of Bitcoin or other big networks.

Risks and Considerations

• Model limits: deterministic, quantized and size-capped models restrict practical AI use cases.
• Adoption: the value of on-chain inference depends on developers actually deploying AI contracts; public usage data is limited.
• Security budget: a smaller proof-of-work network is more exposed to hashrate attacks than large ones.
• Documentation age: much of the technical documentation dates from the 2019 mainnet era; check official channels for the current state of ZkMatrix and client releases.
• Legacy tokens: old ERC-20 CTXC is not the same asset as native CTXC; verify which one an exchange or wallet supports.

Why Cortex Matters

Cortex is a concrete, working example of executing AI inside a blockchain's consensus rules rather than beside it. For anyone comparing AI-crypto projects, it marks one end of the design range: maximum verifiability, with tight limits on model size and complexity.

Where Can You Buy Cortex?

CTXC trades on centralized exchanges. Before buying, confirm that the venue lists native mainnet CTXC and supports deposits and withdrawals on the Cortex network, not the retired ERC-20 version.

Storage Options After Purchase

• Cortex Wallet, the official wallet listed on cortexlabs.ai.
• MetaMask via custom RPC. Cortex uses the same key system as Ethereum, so addresses look like Ethereum addresses; always make sure you are on the Cortex network when sending.
• Keep large balances in self-custody rather than on an exchange, and back up the recovery phrase offline.

Related Coins

• Bittensor (TAO)
• Artificial Superintelligence Alliance (FET)
• SingularityNET
• Ocean Protocol (OCEAN)
• Render (RENDER)