Limits, costs & launch readiness
Understand beta capacity, provider billing, supported features and remaining launch gates.
Beta limits
Each account can create 20 watches, with intervals of 60–86400 seconds. General authenticated traffic is limited to 120 requests/minute, with tighter limits for sensitive actions. Hosted assistant calls have bounded model rounds, tools and output; partial replies are explicitly marked.
Each account supports 20 research routines, with recurring intervals of at least five minutes. Paid social approvals are capped at 20 per day. Proposals expire after 30 minutes. Apify requests use at most five items and a displayed actor-run cap up to $0.25. X endpoint billing is governed by your provider account.
Who pays for what
Remote MCP uses your existing AI client. Optional hosted chat uses a separately billed model API account. Alchemy, Apify and X usage follows the connected provider plan; consumer subscriptions do not automatically cover API usage.
No unlimited-capacity promise or new paid Cloudflare plan is implied. Service limits and public-data provider quotas apply. Pricing for a future Ultron commercial plan has not been announced here.
CoinGecko Demo/Pro requests use the user’s connected account and approved limits. Ultron does not provide an operator-funded paid market-data fallback. Public sources retain shared quotas; bringing an unrelated social key does not lift those limits.
Available versus still missing
Historical research covers up to seven days and 1000 completed candles per pool request; historical leader scans examine at most ten candidates. Wallet history requires a working Alchemy key and supported methods. Profit is calculated only over supplied cost-basis records. FOMO is authorized snapshot context, not a continuous board feed. Host notifications require a separately configured task.
Ultron never executes trades, grants issuer access or starts paid research without approval. Telegram and Discord are coming soon.