Skip to main content

LLM Gateway

A unified interface for LLMs — pay with your launch fees or wallet balance.

Access Claude, Gemini, GPT, Grok, and other frontier models through a single API. Every request is tracked, and you can fund usage directly from your token launch fees or authorize payments in USDC, USDT, ETH, or any other ERC-20 you hold on Base, Polygon, Ethereum, Arbitrum, or BNB Chain.

LLM Dashboard

Manage your LLM Gateway at bankr.bot/llm — view usage breakdown by model, top up credits, and configure auto top-up.

Generate and configure API keys at bankr.bot/api-keys.

Why Bankr LLM Gateway?​

FeatureBenefit
Cost TrackingFull visibility into token usage and costs per request
Launch Fee FundingAutomatically allocate a portion of your token launch fees to AI
Flexible PaymentsPay with USDC/USDT directly or any ERC-20 (auto-swap) on Base, Polygon, Ethereum, Arbitrum, or BNB Chain
Multi-ProviderAccess models from Anthropic, Google, OpenAI, xAI, Z.ai, DeepSeek, MiniMax, Moonshot AI, and Alibaba
High AvailabilityAutomatic failover ensures your agents stay online
SDK CompatibleWorks with OpenAI and Anthropic SDKs — no code changes

Supported Providers​

ProviderModels
AnthropicClaude Fable 5, Opus 4.8, Opus 4.7, Opus 4.6, Sonnet, Haiku
GoogleGemini 3.5/3.1/3/2.5 Pro & Flash, Gemma 4
OpenAIGPT-5.5, GPT-5.4, GPT-5.2, Codex, Mini, Nano
xAIGrok 4.3, Grok 4.20, Grok 4.1 Fast
Z.aiGLM-5.1, GLM-5, GLM-5 Turbo
DeepSeekDeepSeek V4 Pro, V4 Flash, V3.2
MiniMaxM2.7, M2.7 HighSpeed, M2.5
Moonshot AIKimi K2.7 Code, K2.6, K2.5
AlibabaQwen3.7 Plus, Qwen3.6 Flash, Qwen3 Coder, Qwen3.5 Plus/Flash

See Supported Models for the full list with context windows and capabilities.

How It Works​

┌─────────────┐  ┌─────────────┐  ┌─────────────┐
│  OpenClaw   │  │ Claude Code │  │  AI SDK /   │
│             │  │             │  │ Custom Apps │
└──────┬──────┘  └──────┬──────┘  └──────┬──────┘
       │                │                │
       └────────────────┼────────────────┘
                        │
                        ▼
              ┌─────────────────┐
              │   LLM Gateway   │
              │  llm.bankr.bot  │
              └────────┬────────┘
                       │
   ┌──────────┬────────┼────────┬──────────┐
   ▼          ▼        ▼        ▼          ▼
Vertex AI  Vertex AI  OpenAI  MiniMax  OpenRouter
(Claude)   (Gemini)                    (All Models)

Routing Strategy:

  • Claude models → Vertex AI (primary), OpenRouter (fallback)
  • Gemini & Gemma models → Vertex AI (primary), OpenRouter (fallback)
  • GPT models → OpenAI (primary), OpenRouter (fallback)
  • DeepSeek models → OpenRouter (primary)
  • MiniMax models → MiniMax (primary), OpenRouter (fallback)
  • Grok, GLM, Kimi, Qwen models → OpenRouter only

Data Privacy​

When a request is routed through OpenRouter (as a primary route or a failover), the gateway sends data_collection: "deny" on every request. OpenRouter is instructed never to route to upstream hosts that store or train on your prompts.

This governs the OpenRouter hop only — the upstream provider's own policies (for example, Anthropic's for Claude) still apply independently. See each provider's documentation for their data-handling practices.

For a stronger guarantee on supported open-weight models, Private Inference routes a request into a hardware-secured enclave (TEE) by appending :private to the model ID — the gateway verifies the enclave's attestation on every request and never silently downgrades.

Quick Start​

Get Your API Key

Get your LLM Gateway API key →

Base URL:

https://llm.bankr.bot

OpenAI-compatible format (/v1/chat/completions):

curl https://llm.bankr.bot/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "X-API-Key: bk_YOUR_API_KEY" \
  -d '{
    "model": "claude-opus-4.8",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Anthropic-compatible format (/v1/messages):

curl https://llm.bankr.bot/v1/messages \
  -H "Content-Type: application/json" \
  -H "X-API-Key: bk_YOUR_API_KEY" \
  -d '{
    "model": "claude-opus-4.8",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Both formats work with all models. See API Reference for full details.

Key Permissions

Your API key must have LLM Gateway enabled at bankr.bot/api-keys.

  • The Read Only toggle only affects the Agent API — LLM Gateway access is always available when enabled.
  • A key used only for LLM Gateway doesn't need Agent API enabled.

See Access Control for full details.

LLM Credits Required

New accounts start with $0 credits. Requests will be rejected (402) until you top up. Add credits at bankr.bot/llm?tab=credits or via bankr llm credits add <amount>. LLM credits are separate from your trading wallet — see the OpenClaw setup guide for details.

Model Discounts​

The LLM Gateway supports time-bounded discounts on individual models. When a discount is active and you're eligible, your credit deduction uses the discounted price automatically — no action needed.

Check bankr.bot/llm for active promos and your effective pricing. The /v1/models endpoint only returns discounts you qualify for; the discount field on each model includes bps, valid_from, valid_to, reason, and eligibility_scope.

Setup with Bankr CLI​

The Bankr CLI can configure your tools automatically:

npm install -g @bankr/cli
bankr login
bankr llm setup openclaw --install   # OpenClaw
bankr llm setup opencode --install   # OpenCode
bankr llm setup claude               # Claude Code
bankr llm setup cursor               # Cursor

If your LLM gateway key differs from your Bankr API key, you can set them separately:

bankr login --llm-key YOUR_LLM_KEY
# or
bankr config set llmKey YOUR_LLM_KEY

See Coding Agents for full details on each tool.

Credit Management​

bankr llm credits                   # Check balance
bankr llm credits add 25            # Add $25 credits (defaults to Base USDC)
bankr llm credits add 25 --token USDT  # Pay in USDT on whichever chain holds the most
bankr llm credits add 50 --token 0x... --yes  # By contract address, no confirm
bankr llm credits auto              # View auto top-up config
bankr llm credits auto --enable --amount 25 --threshold 5 --tokens USDC,USDT
bankr llm credits auto --disable

Tokens are resolved across every supported chain (Base, Polygon, Ethereum, Arbitrum, BNB Chain) — the highest USD balance wins. USDC/USDT are paid directly when accepted on that chain; anything else is swapped to the chain's preferred stablecoin.

Next Steps​

  • Private Inference — Run select models in a verified TEE with :private
  • OpenClaw — Configure your OpenClaw agent
  • Coding Agents — Claude Code, Codex, Cursor, OpenCode, and more
  • Apps — Integrate via Chat Completions, Anthropic Messages, or the Vercel AI SDK
  • Install the Bankr Skill — Give your skills-compatible agent crypto and DeFi capabilities
  • API Reference — Full endpoint documentation
  • Supported Models — Full model list with context windows and capabilities