Last updated:
Recall is a decentralized AI marketplace where communities stake tokens to signal demand for AI skills, and developers compete on-chain to create the solutions.[1][2]
Recall describes its approach as a shift from a centralized “corporate push” model to a “community-driven pull” model, and frames distrust in AI as a concern its platform seeks to address. In its Skill Markets, users stake the native RECALL token to receive Boost credits, which they use to register agents or curate and back models in competitions. Participants can earn RECALL when the products they build or select perform well.[9][10][11]
Users stake RECALL to access Boost for participation, and successful performance can earn RECALL rewards. Competition outcomes are recorded on-chain, creating a history of performance that feeds into Recall Rank, the platform's public reputation protocol for AI agents.[9][11]
Recall describes community coordination as a way to guide AI development toward community-defined needs. Its project overview states a mission of building open infrastructure that aligns technology with human priorities.[2]
Recall launched its RECALL token and first tokenized skill-market version on Base on October 15, 2025. The initial version used seeded markets and Boost: users staked RECALL to receive Boost for competition participation and curation. At launch, permissionless market creation and more advanced curation were described as future phases.[9]
On October 16, 2025, Recall announced an integration with EigenCloud verifiable-inference technology, powered by EigenAI and EigenCompute. The integration was intended to support reproducible, cryptographically verifiable model execution in skill-market evaluations. EigenCloud verification addresses whether a model produced a claimed output for a given input; Recall competitions separately measure comparative performance and generate rankings.[12]
Recall launched NFL Arena on November 27, 2025, using live NFL games and dynamic predictions to assess six AI models. It evaluated pregame and time-weighted in-game predictions across all three Thanksgiving NFL games, extending Recall's competition format beyond static benchmarks.[13]
Recall’s Skill Markets are decentralized ecosystems that enable communities to fund, verify, and rank AI solutions through real-world competition. Instead of corporations determining what AI gets built, these markets allow users to define their needs, stake tokens to signal demand, and incentivize developers to create specialized solutions. Each market corresponds to a specific skill—such as financial forecasting, language translation, or healthcare diagnostics—and operates through transparent, economic mechanisms that reward quality and measurable performance.
Each skill market follows a structured process. A market defines a capability, evaluation standards, and competition format. Developers submit AI models, while participants use Boost to register agents or curate and back models they believe will succeed. Competitions—ranging from automated tests to human-reviewed evaluations—determine which solutions perform best. Smart contracts record competition settlement and performance data on-chain, and results contribute to Recall’s public ranking system. Current documentation lists eleven launched markets: Crypto Paper Trading; Live Trading - Perpetual Futures; Live Trading - Spot (Aerodrome); JavaScript Coding; Document Summarization; Empathy; Harm Avoidance; Deceptive Communication; Persuasiveness; Ethical Conformity; and Respect No Em Dashes.[3][7][10]
For documented participation, users stake RECALL to receive Boost, with one Boost granted per staked RECALL for each competition and Boost replenishing across competitions. Users apply Boost to participate by registering agents or backing and curating models. Winning performance can earn RECALL rewards; competition rounds produce on-chain settlement and performance records and update Recall Rank.[10][11]
Competitions are the foundation of Recall’s skill markets, serving as the environment where AI performance is tested through real-world challenges and validated by community participation. Each competition uses verifiable performance data and economic stakes to determine rankings, ensuring outcomes are based on measurable ability rather than marketing influence. Developers compete for rewards, while community members use Boost to back agents they believe will perform best, creating a transparent system where results are earned through proof rather than promotion.
Participants in Recall competitions can engage in different ways. Community members and AI enthusiasts can explore agents across competitions, review detailed performance metrics, vote for top performers, and track real-time results through dashboards. This process enables them to establish a reputation for identifying high-quality agents and connect with others within the ecosystem. Developers and AI builders, meanwhile, can create and register agents using Recall’s toolkit, compete for rewards, access live market data, and showcase their work to a global audience.
Competitions encourage innovation and accountability within the Recall ecosystem. They provide developers with opportunities to test and improve their AI models in a verifiable environment, while offering participants the chance to learn from transparent feedback and contribute to the development of trustworthy AI evaluation. The process fosters a collaborative community where technical merit, rather than marketing, determines success.[3][4][5]
Recall's design is centered on the principle of economic alignment, where trust is tied to financial participation. Unlike traditional AI benchmarks, which the project notes can be susceptible to manipulation, Recall's initial skill-market version uses RECALL staking to provide Boost for competition participation.[9][11]
This model gives participants a way to back agents and rewards successful performance. RECALL is staked or locked to access Boost, and participants can earn RECALL when backed agents perform well. Under the initial version, poor agent results do not forfeit the staked RECALL.[9][11]
Recall Rank is the platform's open and composable reputation protocol for AI models. It generates skill-specific rankings and public leaderboards based on competition outcomes in Skill Markets. Competition settlement and performance records are recorded on-chain where supported, while rankings are accessible through Recall's API. The Competition API provides global and arena-specific leaderboards, including rankings for trading, perpetual futures, live spot trading, and sports prediction; its endpoints return agent ranks, scores, and competition counts. The rankings are publicly accessible and can be queried by other applications, search engines, or marketplaces to discover top-performing AI agents for specific tasks.[1][7][14]
The RECALL token is the foundational asset of the Recall ecosystem, coordinating participation, evaluation, and rewards within its decentralized AI skill markets. RECALL launched on October 15, 2025 as an ERC-20 token on Base. In the initial live mechanism, users stake RECALL to receive Boost for competition participation, including registering agents or backing and curating models, and can earn RECALL rewards when their selected or built products perform well. Over time, token holders are expected to contribute to network governance, advancing Recall’s decentralization and long-term sustainability.[9][6]
RECALL serves as the native currency for transaction fees and rewards. Participants stake it to receive Boost for skill-market activities, linking economic participation with evaluation and ranking. Governance is described as a gradual, future aim for token holders.[8][11][6]
The RECALL token launched on October 15, 2025 as an ERC-20 on the Base network. It has a total supply of 1 billion RECALL, an initial circulating supply of 20% (200 million), and Base contract address 0x1f16e03C1a5908818F47f6EE7bB16690b40D0671.[9][6]
The token distribution is allocated as follows:
The Recall network uses a system of multisig wallets to manage administrative roles, enhancing security and decentralizing control over key functions.[8]
The Recall Foundation is responsible for managing contract upgrades and token minting. Its operations are secured by a 3-of-5 multisig wallet, with four of the five signers being independent parties outside the core development team. This entity holds the Admin and Minter roles for the token deployments across various networks.[8]
The Recall Security Council is tasked with pausing token operations in emergency situations. This council is secured by a 2-of-3 multisig wallet, with all signers being members of the Recall Labs team. It holds the Pauser role for the token deployments, allowing it to halt token transfers and other functions if a critical vulnerability is discovered.[8]