On 15 June 2026, Wallet V, a self‑custody Web3 wallet, rolled out the first public performance benchmark for the AI trading agents its users have built on the decentralized derivatives platforms Hyperliquid and Aster. The benchmark, live on Wallet V’s website, aggregates the on‑platform results of 688 agents that were created by Wallet V users over the past two months.

Each agent was assembled by a user who chose a large‑language model (LLM) to generate trading signals. The models then executed perpetual futures orders on Hyperliquid or Aster across a spectrum of asset classes. Wallet V groups the agents by the underlying LLM family and reports aggregate cohort performance. The data set spans seven distinct LLM families, and the benchmark refreshes automatically whenever new agents are deployed.

Overall, 42 percent of the agents recorded a profit and loss (P&L) balance of zero or higher during the measurement period. The highest agent‑level return on investment (ROI) observed in the cohort was +307 percent, while the lowest was –30 percent. Models represented by fewer than ten agents are reported only directionally, as the sample size is insufficient for statistical significance.

The agents traded perpetual futures across four broad asset classes available on Hyperliquid and Aster. These include major digital assets such as Bitcoin (BTC), Ethereum (ETH), and Solana (SOL); pre‑initial public offering (IPO) equity exposure; commodities such as gold, silver, and oil benchmarks; and major foreign exchange (FX) pairs. All instruments were accessed through third‑party venues, and the agents’ trades were carried out on the respective decentralized platforms.

By making performance data publicly available, Wallet V provides a transparent view of how different LLMs perform in live market conditions, allowing users to evaluate models in a manner similar to how institutional investors assess portfolio managers. The release signals a broader industry shift toward quantifiable metrics for algorithmic strategies that operate on decentralized infrastructure.

Wallet V plans to expand the benchmark in future releases. Upcoming features will include support for newer LLM families, integration of prediction markets, and advanced analytics tools that enable copilot trading and personalized AI prompt generation tailored to individual trading styles. The company said these additions will further enhance users’ ability to customize and monitor their agents.

Wallet V is an incubation project of Virgo Group, a digital asset service provider led by CEO Adam Cai. The wallet offers users access to third‑party AI models for configuring trading agents and executing user‑defined strategies. Virgo Group is backed by investors such as Draper Dragon, OKX Ventures, Vaulta Foundation, Cobo Ventures, Waterdrip Capital, and Sora Ventures. Wallet V’s mobile applications for iOS and Android are available at dl.walletv.io.

The company clarifies that Wallet V is a software provider that connects to external platforms and does not offer trading services or AI automation tools directly. It does not provide investment, tax, or legal advice, and access to certain products may be restricted in some jurisdictions. Users are advised to conduct their own due diligence before engaging in high‑risk trading activities.

In summary, Wallet V’s public benchmark marks the first systematic aggregation of AI agent performance on Hyperliquid and Aster. The data set offers a snapshot of how different LLMs fare in live perpetual futures markets and sets a precedent for transparency in decentralized AI‑driven trading.