Skip to content
⌘K
InboxAsk a human

LLM Gateway & Credits

Updated by Frenchie

In this article

In this article group

LLM credits are what you spend when you use Max Mode in chat or call AI models through Bankr's LLM Gateway. To add credits, ask Bankr "add $25 in LLM credits" or top up at bankr.bot/llm. The gateway is one API for models from many AI providers, and your agent can pay for it with the trading fees its token earns.

To turn on Max Mode in chat, or to compare it with a Bankr Club subscription, see Bankr Club & Max Mode.

Managing Credits

In chat:

"add $25 in LLM credits"
"what's my credit balance?"

Web: bankr.bot/llm?tab=credits

CLI:

bankr llm credits           # check balance
bankr llm credits add 25    # add $25
bankr llm credits auto --enable   # auto top-up

Available Models

Max Mode gives you premium models from Anthropic (Claude), OpenAI (GPT), Google (Gemini), xAI (Grok), Moonshot, Alibaba, and more. Outside Max Mode, the agent runs on a fast base model. The current lineup and prices are always at bankr.bot/llm, since models are added, discounted, and retired often.

  • Max Mode on the web and in the CLI offers the newest premium models. Lighter models are available through the gateway API only.

  • New models, price changes, and removals reach Max Mode within about a minute.

  • If your saved Max Mode model is retired, Bankr switches you to its replacement. If there isn't one, Max Mode turns off.

  • If a model is withdrawn mid-conversation, that run is billed at the model's last published rate.

  • If the CLI rejects a --model value, it lists the accepted IDs. Run bankr update if a new model isn't accepted yet, and bankr llm models to see current models.

Model Discounts

New models often launch with intro pricing, and existing models get occasional promo rates, each for a limited time. Your actual price, including any discount, is shown live at bankr.bot/llm.

What Happens When Credits Run Out?

Bankr checks your effective balance: your credit minus usage that's been counted but not yet settled. A run that just finished can still be pending, so your effective balance can be lower than the number you see.

  • Max Mode in chat: a new message gets a top-up prompt instead of an answer (on X it falls back to the base model). A message already running stops when it reaches your balance and tells you what it used and which transactions it already sent. Top up and ask Bankr to continue; your progress stays in the thread.

  • Automations: skipped runs show in the automation's history, and 3 credit failures in a row pause the automation.

  • Gateway API requests: rejected once your effective balance hits zero.

Usage your balance can't cover stays owed and is taken from your next top-up. Turn on auto top-up to avoid running dry mid-task.

If your top-up payment went through on-chain but your credit balance hasn't changed after a few minutes, open a ticket through the Help Center form with the transaction hash, the chain, the amount you paid, and your login handle or email.

Building Against the Gateway API?

The gateway works with OpenAI and Anthropic SDKs. Prompt caching works on Claude models, and cached reads are billed at a reduced rate. Endpoints, keys, and integration details are in the LLM Gateway docs.

For more detail, see LLM Gateway, Max Mode, and Credit Management on docs.bankr.bot.