What VTX does, what it costs, and how to run your own AI trading bot.
VTX Macro is a platform for running AI trading bots on Hyperliquid. You choose the model, markets, prompts, and settings. The AI analyzes the information it receives and makes trading decisions; VTX handles the connection to the exchange and execution. You can inspect other bots and their results on the Leaderboard.
No. The platform is already built. You configure a bot through the app, and you can copy an existing bot's settings as a starting point. Coding tools and agent integrations are optional. Start offers guided setup, agent-assisted setup, and a manual path.
Create a VTX account, connect your Hyperliquid wallet through its trading API, and configure access to an AI model. Hyperliquid must use the supported Manual account mode, with collateral in the appropriate Perps balance. Follow Start or Help for the connection steps, then inspect the bot's analysis with AI Assistant before enabling AI Trader.
Yes. Open a bot's configuration from the Leaderboard to inspect its prompts, models, markets, and settings, then use Copy Settings for your own profile. You still need your own exchange connection and model access. Copying settings creates your own configuration; following another wallet's orders is the separate Copy Trading feature.
Begin with Start for the guided setup options and Help for detailed instructions. Use Join Telegram at the top of this page for community help. For product support, use Contact and include your profile handle and a description of the problem. Never send passwords, API secrets, private keys, or seed phrases.
Client Mode is free. Optional Server Mode costs $0.0025 per qualifying model call and $0.25 per filled exchange order. Each filled order counts once; opening and closing are separate orders.
AI provider charges, exchange trading fees, and funding are separate. A free AI model can avoid provider token charges within its free-tier allowance.
In Client Mode, your browser, desktop app, or headless runtime runs the bot and must stay online. Server Mode runs on VTX infrastructure and can continue after you close the app. You can enable optional Server Mode Backup to take over an already-running client bot when its client becomes unavailable, subject to server credentials, model availability, capacity, and fees.
It depends on the provider, model, prompt size, analysis frequency, and any additional Review or Screener calls. Bring your own key (BYOK) means provider usage is billed to your provider account. The Models page shows available token prices. Prompt cost estimates describe model usage; they do not include VTX's separate Server Mode fees or show a subscription's remaining allowance.
You can choose from the live Models catalog, supported local models, or connected providers and agents. Compare cost, available context, response time, and how the model behaves with your strategy. A more expensive model is not a guarantee of better trading results. Public bots provide examples of settings and outcomes you can inspect.
Yes, when your provider offers a free model or free-tier access. VTX's model selectors include a confirmed-free-model filter, and Help includes a Google AI Studio setup guide. Free access still has provider limits. More symbols, larger prompts, shorter intervals, or extra model calls can exceed those limits; availability and allowances are controlled by the provider.
Yes. Client Mode supports Local AI through compatible servers such as LM Studio and Ollama using an OpenAI-compatible connection. Configure the server in System, test the connection, then select the local model on AI. Your hardware must support the model and its prompt size. The local server and client runtime must remain reachable; trading still requires an internet connection.
Yes, through supported connections. These include subscription-backed inference such as Codex and provider-specific coding plans, plus agent-driven integrations. Capabilities differ: some connections automatically answer VTX prompts, while others let an agent request data and submit decisions. A chat or coding subscription does not automatically include API access. Use the connection instructions in Help to check your service and setup.
The model receives your instructions and trading context, then returns a decision and an explanation. It can choose to buy, sell, or hold and specify trade size within the configured limits. AI Trader can execute accepted decisions on Hyperliquid. AI Assistant runs the analysis without executing trades.
The prompt can include candles, indicators, market statistics, account balances, positions, performance, economic-calendar information, and news. You choose the available variables and instructions on the AI page. Trader prompts also receive recent decision and position context. Inspect the recorded execution context to see what was actually sent for a particular decision.
An analysis call does not have to produce a trade. The model can keep returning HOLD while it waits or manages an existing position. In the normal model workflow, the Loop Interval controls how often analysis runs; Review and Screener can add calls. A connected Main Agent can choose its own cadence. Prompt count measures analysis activity, not trade count.
No. Each profile can use different models, symbols, timeframes, prompts, and limits. The system prompt describes how you want the model to trade; the user prompt supplies its context. You can copy a configuration or experiment with your own. The model interprets these instructions, so a sentence in a prompt is not a guaranteed execution rule.
Yes. The AI page includes trade-size and exposure limits, optional mechanical stop-loss and take-profit settings, scheduling, and other risk controls. Some settings enforce application limits; others guide the model through its prompt. Review the setting's behavior in Help. Stops and limits cannot guarantee a fill price or prevent every loss, and some controls need an active runtime.
Yes. Public profiles expose bot settings, prompts, decisions, trades, and results. Trade Analysis shows the recorded decision and available execution context; provider reasoning traces appear when the provider supplied them and VTX retained them. The public Changelog records settings changes, which helps you distinguish today's configuration from the one used for an older result.
No. The application is closed source. Its trading configurations, prompts, decisions, and results are public so people can inspect how bots are configured and learn from their outcomes. Public strategy information does not mean the application's source code is available.
VTX is built around public bot configurations and trading transparency. Treat strategy settings, prompts, and recorded model output as public, and do not put secrets in your prompts. Account login details, billing information, provider credentials, and exchange secrets are excluded from public profile views. See Privacy for the data-sharing details.
No. Your bot starts at a different time with its own balance, positions, runtime, and model access. Market conditions and execution can differ, and the same model can make different decisions. A copied configuration is a starting point to inspect and evaluate; it does not copy the other bot's trading history or guarantee its future returns.
Your trading funds stay in your Hyperliquid account. VTX connects using your main wallet's public address and the separate API wallet signing key you authorize for trading. Do not enter your main wallet's private key or recovery phrase. Trading authority still carries risk: the bot can place losing trades. Follow the API wallet setup in Help.
AI Trader places real orders on connected Hyperliquid accounts. The Leaderboard and profile analytics report live-account trading results, not simulated paper portfolios. Open a profile to inspect its trades, selected period, and settings history rather than relying on a screenshot alone.
Use AI Assistant on the Trade page to run the analysis and inspect what the model would decide without executing trades. This is a dry run: it does not maintain a simulated portfolio or generate paper-trading PnL. Model calls can still consume provider allowance or incur the applicable provider and Server Mode fees.
No. Volume is the notional value traded. The same funds can be traded repeatedly, so a small account can accumulate a large volume while making a profit or a loss. Use PnL and ROI for the selected period to examine performance, and distinguish realized results from the changing value of open positions.
No. Results depend on the market, symbols, model, settings, costs, and execution. A strategy that benefits from a strong trend can struggle in a sideways market or reversal. Compare different periods, losses, and configuration changes as well as gains. AI can make mistakes and lose money; past results do not establish future profitability.
VTX supports available Hyperliquid perpetual markets, including HIP-3 markets that can reference stocks, indices, or commodities. These are perpetual derivatives, not ownership of company shares or physical assets. Check the live symbol selector for availability. Different markets can use separate collateral balances, which you fund through Hyperliquid; VTX does not transfer that collateral for you.