Just grabbed 20 Slonks floor tokens on OpenSea over the past couple days. This thing launched May 1st and honestly, the momentum is wild—went from mint to 0.062 ETH in 5 days, with 318 ETH moving in the last 24 hours. Let me break down what's actually happening here, because the design is genuinely clever.



So the basic idea: take 10,000 CryptoPunks, feed them into a tiny transformer model embedded directly into an Ethereum smart contract, and let the AI redraw them. Yeah, you read that right—there's an actual neural network running on-chain. The genius part? They leaned into the flaws. AI art gets roasted for looking like AI, right? They made that the entire mechanic. The more errors the model makes, the higher the score. They literally called it "Slop is the art" and made it the project name.

Each NFT has three key attributes. First is the punk number—which of the original 10,000 it's mimicking. Second is the slop value—basically how many pixels the AI messed up out of 576 total. A card with 287 slops means 287 pixels were drawn wrong. This number drives everything. Third is level, which starts at 0 and increases with each merge.

Here's where it gets interesting. You can merge two cards of the same level—burn one, level up the other. What happens: the AI averages the two "impressions" and redraws. The result? Definitely less similar to the original punk. Lower similarity equals higher slop score. That slop converts to $SLOP tokens, which you can sell. The trade-off is permanent—the burned card is gone forever.

Then there's the token mechanic. You can disassemble any NFT for $SLOP tokens based on its slop value. A 287-slop card gets you 287 tokens. This transforms an indivisible art piece into divisible, tradeable tokens. If a card sells for 0.06 ETH on the market but the tokens inside are worth 0.10 ETH on Uniswap, you arbitrage it. Higher slop means more token value—it's got better "coin content."

The third option is the lottery. Burn tokens to spin for a new card pulled from previous dismantled ones, redrawn by the AI with added noise. 50% chance it's clean with low slop, 49% chance it's mixed, and 1% chance the AI goes absolutely haywire—near-perfect slop that looks like complete chaos. That 1% is the jackpot. It's a Dutch auction starting at 576 tokens, dropping by 1 every few minutes, floor of 100. But if someone buys, it resets to 576. Early is expensive, late is risky—someone might snatch it first.

What really got me though is the engineering. They actually put a neural network on-chain. The 22.7 KB transformer has its weights split across 9 contracts using SSTORE2 storage. Every mint or merge runs forward inference in the EVM, drawing the SVG on the spot. 10-dimensional embeddings, 18 attention heads, 10,000 vocabulary. It's a tiny model but it's genuinely working.

Total token supply is 5,760,000—exactly 10,000 times 576. When you send an NFT into the void, the contract counts the defective pixels and mints you that many tokens. Burn tokens to pull a new card from the void, redrawn with noise added. I haven't seen any other project doing pixel-level NFT tokenization like this.

The design is internally consistent. NFT collectors can collect by preference—some chase rare original types, others go for heavily merged hybrids. Creating generative art on-chain eliminates IPFS dependency entirely. GameFi players get arbitrage puzzles and the satisfaction of optimization. Different people, different reasons to engage.

There's real technological innovation here, not just hype. The learning curve is steep, yeah, but good games usually are. Once people get it, you'll probably see a solid community form. Blockchain innovation like this deserves attention.
ETH-1.02%
UNI-2.68%
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pin