Which AI powers your trades
BrokerBridge does not include its own AI. You bring one: a provider account like Claude or ChatGPT, your own API key, a model running on your own computer through Ollama, or BrokerBridge credits if you would rather not manage a provider account yourself. Whichever you pick, it reads market data, the rules you set, and your current account context, which includes your open positions, your profit and loss for the day, and how much of your account is in the market. It never decides how you approve a trade; that choice is separate, and you set it yourself.
Supported providers include Anthropic (Claude Sonnet/Opus 5), OpenAI (GPT-5.6), Google (Gemini 3.5), xAI (Grok 4.5), and DeepSeek v4 or other open-source models through OpenRouter. Whichever one you choose, you still pick I approve every trade, I confirm from the alert, or I approve the strategy up front per strategy, and the AI Arena uses its own practice-money track.
Provider comparison
| Provider | Route | Auth | Billing source | Availability |
|---|---|---|---|---|
| Anthropic (Claude) | Direct provider | OAuth or API key | Selected Anthropic account | When configured |
| OpenAI | Direct provider | OAuth or API key | Selected OpenAI account | When configured |
| OpenRouter | Direct provider | API key | OpenRouter usage | When configured |
| Google AI | Direct provider | API key | Google AI usage | When configured |
| xAI (Grok) | Direct provider | API key | xAI usage | When configured |
| Ollama (local) | Local loopback | No provider API key | Your local compute | When Ollama is running |
| BrokerBridge managed credits | Managed model plane | Verified entitlement and wallet | BrokerBridge credits | Pay-as-you-go, from the dashboard |
BrokerBridge-managed credits
Managed credits are an optional pay-as-you-go AI route. A wallet in the dashboard reserves an authorized amount before a model turn and settles recorded usage afterward. Running out of credits blocks the managed turn rather than switching to another billing source.
Credits pay for the complete BrokerBridge-managed operation, meaning model inference, metered tools, infrastructure, and service margin. They are not a resale of raw provider tokens.
| Pack | Credits | Price |
|---|---|---|
| $10 pack | 1,000 credits | $10 |
| $20 pack | 2,000 credits | $20 |
| $50 pack | 5,000 credits | $50 |
| $250 pack | 25,000 credits | $250 |
1 credit = $0.01. You can also enter a custom amount from $10 to $2,000 instead of a listed pack; it buys credits at the same $0.01 rate. A card is required only when you choose this managed-credit route, through Stripe-hosted checkout on an explicit first pack purchase, never implied by, or bundled into, a subscription. Your card is saved on any credit purchase, whether a listed pack or a custom amount, so that a future automatic top-up would not need you to re-enter it. Auto-refill itself is not available yet: there is no control to turn it on, so no top-up is charged automatically. When it does ship, switching it on will require explicit consent covering the refill amount, the balance threshold that triggers it, a monthly cap, and a one-tap revocation control. A zero credit balance blocks managed-model work only, never your account, the desktop app, local Ollama models, or your broker connection.
Buying credits: purchase a pack from the dashboard whenever you need one, no separate approval step. Every purchase is a deliberate action you take; nothing is charged on a schedule.
Run a local model with Ollama
Ollama is optional. It runs the AI model on your own computer, so nothing you type leaves your machine and local inference does not require a provider API key. BrokerBridge looks for Ollama at http://127.0.0.1:11434 and shows the chat-capable models you have installed.
- 1.Install and start Ollama on the same computer as BrokerBridge.
- 2.Download a chat model with Ollama. Choose its size for the memory, disk, and response speed available on your computer.
- 3.In BrokerBridge setup, select Ollama (local), then select one of the detected installed models.
- 4.Try the selected model with practice money before relying on it for any other workflow.
Hardware guidance: model capacity is not one-size-fits-all. Apple Silicon uses Metal acceleration; supported NVIDIA and AMD GPUs can accelerate Ollama; CPU-only use is possible but may be slower. Model files can require tens to hundreds of gigabytes, and larger context windows need more available memory. Check Ollama hardware support and macOS requirements for your machine before downloading a model.
Anthropic (Claude)
Claude is one supported provider. Model behavior, availability, and billing depend on the account and model selected in Settings.
OAuth (use your existing subscription)
If you already have a Claude subscription (Pro, Team, or Enterprise), you can authenticate via OAuth to use your existing subscription. The setup wizard guides you through the sign-in flow.
API key (bring your own)
- 1.Go to console.anthropic.com and create an API key.
- 2.In the setup wizard, select "Anthropic" as your provider and "API Key" as auth type.
- 3.Paste your key. It is encrypted locally and never sent to BrokerBridge servers.
OpenAI
OpenAI is available through the supported subscription-authentication route or a usage-billed Platform API key. Availability and billing follow the route selected in Settings.
- 1.Go to platform.openai.com/api-keys and create a key.
- 2.Select "OpenAI" in the setup wizard and paste your key.
OpenRouter
OpenRouter, one bill for many AI models, provides access to many models through a single API, including DeepSeek v4 and other open-source models. Useful if you want to switch between different models without changing provider configuration.
- 1.Go to openrouter.ai/keys and create an API key.
- 2.Choose a model from the OpenRouter model list. Set it in your config.
Google AI
Google AI provides Gemini models through AI Studio. Model availability, latency, and price depend on the selected account and current provider terms.
- 1.Go to aistudio.google.com/app/apikey and create an API key.
- 2.Select "Google" in the setup wizard and paste your key.
xAI (Grok)
xAI provides Grok models through the xAI API. Model availability, latency, and price depend on the selected account and current provider terms.
- 1.Go to console.x.ai and create an API key.
- 2.Select "xAI" in the setup wizard and paste your key.
How AI decisions work
When BrokerBridge identifies a trade setup, it sends the AI a structured prompt containing:
- •Price action data (recent OHLCV bars and extracted features)
- •Evidence snapshots from data source plugins (scanner signals, flow data)
- •Current portfolio state (open positions, daily P&L, exposure)
- •Your rules (the limits you configured)
The AI returns a structured decision: approve or reject, with a confidence score (0-1), reasoning summary, suggested position size, and risk/reward ratio. This decision is recorded in the decisions CSV and presented alongside the original trade plan.
Important:Model output alone doesn't place a trade. Supervised strategies need your approval on each trade plan. I confirm from the alert strategies need a tap within the time window before the setup expires. I approve the strategy up front strategies need your standing approval and a passed review, and only trade inside their configured money limit, your rules, allowed instruments, schedule, and broker mode. The continuous Arena competition may run repeatedly, but only with practice money.