New AI trading co‑pilots arrive

Crypto trading platforms keep trying to turn chatbots into brokers. This week, three of them pushed that idea further. Mode Network promoted a new version of its AI trading agents with zero trading fees. ChainGPT rolled out AI Hub V2 as a bundle of tools for trading, contract analysis, and onchain work. Binance, the biggest exchange in crypto, launched a paid beta called Ai Pro that can analyze markets, place trades, and watch a portfolio from inside the Binance app and website (paragraph.com) (chaingpt.org) (binance.com). That cluster of launches matters because these products are not pitched as better charts or faster alerts. They are pitched as co-pilots. The promise is simple: tell the system what you want, let it interpret the market, and let software do more of the work. Binance says Ai Pro is a “one-stop” AI agent built on the OpenClaw ecosystem. It creates a virtual sub-account, binds an API key to it, and can handle spot orders, perpetual futures orders, leveraged borrowing, price analysis, wallet lookups, and custom strategy execution (binance.com 1) (binance.com 2). That is a bigger shift than the marketing language makes it sound. Binance is not just adding an assistant that answers questions. It is wrapping several large language models into a trading interface and charging for the privilege. In its own guide, Binance says users can plug in models such as ChatGPT, Claude, Qwen, MiniMax, and Kimi. During the beta, activation is priced at $9.99 a month, rising to $29.99 a month after that. The service first became available on March 25, 2026, with activation initially limited on mobile to Android (binance.com 1) (binance.com 2). Mode is taking a different route. Its V4 trading agents are not framed as open-ended chat companions. They are rule-based systems that all read the same market data and the same indicator stack, then let different models interpret those signals. Mode says V4 adds gold and silver alongside bitcoin, uses Donchian Channels and MACD as the fixed strategy core, and lets traders either follow an agent’s call or invert it through a “Counter Mode.” The company’s own write-up says the point is not prediction but transparent, deterministic execution with user-controlled automation (paragraph.com) (mode.network). That design is revealing. Mode is trying to solve the central problem with AI trading tools by shrinking the AI’s freedom. The models do not invent strategies from scratch. They sit on top of a pre-set framework. Binance is looser. It offers broader capabilities and more model choice, but its own materials describe Ai Pro as a semi-automated co-pilot, not a fully autonomous trader. That caveat is doing real work. These systems can execute, but the platforms still need users to believe there is a human somewhere in the loop (paragraph.com) (binance.com). ChainGPT sits between those two poles. AI Hub V2 is less a single trading robot than a Web3 control panel. ChainGPT describes it as an all-in-one hub for traders, builders, and creators, with tools for price analysis, chart context, smart contract work, and other blockchain tasks. Its roadmap had flagged Crypto AI Hub V2 as one of the company’s major product releases, which makes this week’s push look less like a surprise launch than the public packaging of a broader plan to turn crypto workflows into AI prompts (chaingpt.org) (docs.chaingpt.org). The striking part is not that crypto companies are using AI. That has been true for years. The striking part is how quickly they are moving from analytics to delegated action. Binance now sells an agent that can trade from a segregated sub-account. Mode offers onchain agents that can execute the same strategy across multiple models and even across assets like gold and silver. ChainGPT is bundling trading and contract tooling into one interface. The industry keeps calling these products assistants, but the important word is the one they keep sneaking in beside it: execution (binance.com) (paragraph.com) (chaingpt.org).