Road Town, British Virgin Islands, June 15th, 2026, Chainwire
The benchmark covers 688 agents created by Wallet V users over the prior two months. Each agent was configured by the user, used a large language model selected by the user to generate trading decisions, and executed on Hyperliquid or Aster. Wallet V aggregates the on-platform performance of those agents by underlying model. Performance is refreshed as new agents are deployed.
The cohort spans seven large language model families. Across the cohort, 42 percent of agents recorded a profit and loss balance of zero or higher over the period. Peak agent-level return on investment in the dataset ranged from negative 30 percent on the lowest-performing model to positive 307 percent on the highest. Models represented by fewer than 10 agents in the cohort are reported as directional rather than statistically conclusive.
Agents in the cohort executed strategies as perpetual futures across four asset classes available on Hyperliquid and Aster. These include major digital assets such as BTC, ETH, and SOL; equities, including pre-initial public offering equity exposure; commodities including gold, silver, and oil benchmarks; and major foreign exchange pairs. All instruments are accessed through third-party venues.
Wallet V plans to extend the benchmark in subsequent releases. Future releases include the addition of newer model families, support for prediction markets, advanced analytics features for copilot trading and personalized AI prompt generation tailored to each user’s trading style.
About Wallet V
Disclaimer
Trading crypto, perpetual contracts, tokenized assets, and prediction markets involves significant risk of loss and is offered by third-party platforms. Wallet V is a software provider that connects to external platforms and does not offer trading services or AI automation tools directly or indirectly. Wallet V does not provide investment, tax, or legal advice. Access to certain products may be restricted in some jurisdictions.
Peter Ip
