# Into: CookingTAO CookingTAO wants to make Bittensor mining something you deploy with a click, and it pays the developer who wrote the miner only when that code actually performs. // App store for miners, paid on results --- > New to Bittensor? Start here. Experienced users can skip to the full analysis. ### What is CookingTAO? CookingTAO is subnet 122 on Bittensor, and it runs a marketplace for mining code. Developers (people, or AI agents) submit code that mines on Bittensor subnets, ordinary users deploy that code without the usual technical setup, and the developers earn only when the users running their code achieve measurable results. Bittensor is a network where independent operators called miners compete to do useful AI work and get paid in tokens for it; CookingTAO's pitch is that most people who could build good miners never do, because the setup is hard. **The simple version:** It is like an app store for Bittensor miners, where the developer gets paid only when their app actually wins. **Centralized equivalent:** No direct equivalent. The closest analogues are builder marketplaces like Hugging Face or Kaggle, but with on-chain, pay-for-performance rewards instead of listings or prize pools. **How it works:** - **Miners** are the submitted mining-code deployments. A developer's reward tracks how well the real users running that code rank and earn on the original subnets the code targets, not simply whether the code was uploaded. - **Validators** verify user activity, apply CookingTAO's scoring rules, and submit the on-chain weights that direct emissions. The design has them follow a fixed formula rather than exercise discretion. --- ### Why This Matters - **The problem it solves:** Mining on Bittensor still demands real technical work: wallet management, infrastructure, dependency setup, miner configuration, and constant operations. That gate keeps a large pool of capable AI builders out of the network. - **The opportunity:** There are far more people building models, datasets, and agents on GitHub, Hugging Face, and Kaggle than there are people running Bittensor miners. CookingTAO is aimed squarely at converting that group into productive subnet participants. - **The Bittensor advantage:** Rewards settle on-chain and are tied to measured performance, and the marketplace is coordinated by a subnet whose scoring is meant to be transparent and rule-based rather than a private payout decision. - **Traction signals:** Early, and worth stating plainly. CookingTAO was announced as the first incubation project of Lamida Global, a Bittensor-focused firm, and the team published its SN122 incentive-mechanism whitepaper to its GitHub on June 30, 2026. On-chain, the slot was registered on May 22, 2026 and the marketplace is still bootstrapping: there are no active adopter-miners yet, and the project website was unreachable at the time of writing. --- ## Full Analysis **Category:** Other (Bittensor mining marketplace) | **Centralized Competitor:** No direct equivalent; closest are Hugging Face and Kaggle Bittensor pays for useful intelligence, but each subnet still asks participants to run and maintain real infrastructure. CookingTAO's framing of the gap is direct: a large market of AI builders exists, yet it does not translate into scalable miner participation because the process involves crypto risk, technical gating, heavy infrastructure, and uncertain rewards. Subnet 122 is the coordination layer for a marketplace meant to close that gap. **Mechanism:** The following is CookingTAO's design as set out in its incentive-mechanism whitepaper, published to the project's GitHub on June 30, 2026. It separates three roles: Developers submit verified mining code, Users deploy it, and Validators score and submit weights. The stated core rule is that developers are not paid for uploading code; they are paid when real users deploy that code and it produces measurable performance on the subnets it targets. Scoring, per the whitepaper, weights performance far above adoption: a developer's score leans about 90 percent on how well adopters rank and earn on the original subnet and about 10 percent on raw usage, so that fake or idle deployments contribute little. Individual miner shares are then normalized, and a hard per-miner cap (the document gives 30 percent as an example) prevents any single developer from capturing the pool, with any excess left unallocated. A profitability step scales the total miner pool down when platform earnings do not cover the maximum miner allocation. Two design choices are worth flagging for a Bittensor reader. First, the owner side is a burn path, not a rake: the owner UID receives only the unused portion of the miner allocation, which is treated as burned, so weight that miners do not earn is removed rather than paid out. Second, submitted code is private by design. The whitepaper describes runtime code fetching, restricted visibility, and CookingTAO-managed deployment, so that competing miners cannot read each other's strategies through the platform. That private-code model is the main reason the public GitHub footprint is thin: the whitepaper and an assets repository are visible, but the miner code itself is not meant to be. The on-chain picture matches an early-stage subnet. The alpha trades around 0.01060 TAO for a market cap near 4,600 TAO, with pool depth around 2,171 TAO on the TAO side. The subnet's share of network emissions currently reads 0 percent, and its miner-burn sits at 100 percent. Under Bittensor's price-based emission model, live since June 2026, a subnet's network emission share scales with root_prop times its EMA price times (1 minus miner_burn), so a 100 percent miner-burn drives that share toward zero. That is consistent with CookingTAO's own design: with no adopter-miners yet producing measurable performance, the miner pool goes unused and routes to the owner burn path. The slot is inside its network immunity window, which runs to roughly September 19, 2026. Lamida Global's public announcement framed the incubation with a revenue check-in dated November 15, 2026, which lines up with the whitepaper's plan to sustain the project on platform fees rather than emissions. --- ### Risk Factors - **Two-sided adoption is the whole game:** The model needs both developers submitting code and users deploying it, and rewards only flow once adopters produce measurable results on other subnets. Until that happens, the miner pool stays unused and routes to the owner burn, which is exactly why the on-chain network emission share is near zero today. Bootstrapping both sides of a marketplace at once is hard. - **Deregistration after immunity:** Subnet 122 is protected from deregistration only until roughly September 19, 2026. After that, a subnet sitting at the bottom of the network on EMA price becomes a deregistration candidate. Lamida's own November 15, 2026 revenue check-in sets a public clock on proving traction. - **Early and partly unverifiable public footprint:** The incentive mechanism is a June 30, 2026 draft, the website was unreachable at the time of writing, and the miner code is private by design. That is internally consistent, but it also means the live product is hard to verify from the outside right now. - **Liquidity:** With pool depth near 2,171 TAO, larger swaps will face meaningful slippage. --- Into the next one.