Automatic Failover
Health-aware routing with instant provider switching. If one provider fails, your channels stay online with zero manual intervention.
OpenClaw Integration
OpenClaw + Router AI: AI Agents, Every Channel
Power OpenClaw deployments with health-aware routing, multimodal support, and 275+ models. Automatic failover, cost visibility, enterprise security.
Why Router AI for OpenClaw
Build reliable, cost-effective AI agents across every channel
Multimodal Ready
Vision, audio, PDF, and video support across all channels. Run rich multimedia workflows with WhatsApp, Telegram, Discord, and Slack.
Cost Visibility
Per-channel usage tracking with token-level accounting. See exactly what each bot, channel, and conversation costs in real-time.
Enterprise Security
API key isolation, audit logging, GDPR/CCPA compliance. Zero prompt storage by default. Built for enterprise trust requirements.
Unified API
Single OpenAI-compatible endpoint for 275+ models across Anthropic, OpenAI, Google, Bedrock, xAI. No vendor-specific code.
Direct Connections
Every request hits the real provider's real model. No substitution, no pools, no middlemen. Verify on every single request.
Multimodal Capabilities Across Providers
Run vision, audio, and document workflows across all channels
✓ = Full support • = Partial support — = Not supported
Anthropic OpenAI Google (Vertex AI) Bedrock xAI Text Vision Audio PDF Video
Anthropic OpenAI Google (Vertex AI) Bedrock xAI
Power AI Across Every Channel
Build once, deploy everywhere
Rich multimedia support for business conversations
Telegram
Fast, secure bots with inline queries and webhooks
Discord
Community engagement with slash commands and threads
Slack
Workplace automation and team collaboration
China's leading messaging platform for business and social
Enterprise Security & Trust
Built for production deployments with compliance, audit trails, and privacy guarantees
KMS Envelope Encryption
BYOK credentials protected by AWS KMS with per-credential data keys, AES-256-GCM, and hardware-backed master keys.
Audit Logging
Complete request history with channel, model, cost, and latency tracking.
GDPR/CCPA Ready
Data retention controls, export API, and right-to-deletion support.
Zero Prompt Storage
Request bodies pass through but aren't persisted by default. Metadata only.
Get Started in Minutes
Three steps to production-ready OpenClaw deployment
Or Install Router AI Skill
Works with Claude Code, OpenClaw, Cursor, Gemini CLI, and 30+ agent tools clawhub install therouter-api npx @skill-hub/cli install therouter-api
openclaw onboard --non-interactive \
--provider-name=therouter \
--provider-type=openai \
--provider-url=https://api.routerai.lol/v1 \
--provider-api-key=${THEROUTER_API_KEY} \
--default-model=anthropic/claude-sonnet-4.6{
// Router AI OpenClaw Configuration
// 35 curated models covering multimodal, reasoning, coding, and cost-effective use cases
// Copy this entire block into your OpenClaw config file
agents: {
defaults: {
model: {
// Primary model for all agents
primary: "anthropic/claude-sonnet-4.6",
// Automatic fallback models if primary is unavailable
fallbacks: [
"openai/gpt-5.2",
"google/gemini-3.5-flash"
],
},
},
},
providers: [
{
name: "therouter",
type: "openai",
url: "https://api.routerai.lol/v1",
// Set your API key via environment variable
apiKey: "${THEROUTER_API_KEY}",
models: [
{
"id": "anthropic/claude-opus-4.6",
"name": "Claude Opus 4.6",
"context_length": 1000000,
"max_completion_tokens": 128000,
"pricing": {
"input": 0.000005,
"output": 0.000025,
"cache_read": 5e-7
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"long_context",
"prompt_caching"
]
},
{
"id": "anthropic/claude-sonnet-4.6",
"name": "Claude Sonnet 4.6",
"context_length": 1000000,
"max_completion_tokens": 64000,
"pricing": {
"input": 0.000003,
"output": 0.000015,
"cache_read": 3e-7
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"long_context",
"prompt_caching"
]
},
{
"id": "anthropic/claude-haiku-4.5",
"name": "Claude Haiku 4.5",
"context_length": 200000,
"max_completion_tokens": 8192,
"pricing": {
"input": 0.000001,
"output": 0.000005,
"cache_read": 1.0000000000000001e-7
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"tools",
"cost_effective",
"prompt_caching"
]
},
{
"id": "anthropic/claude-sonnet-3.7",
"name": "Claude Sonnet 3.7",
"context_length": 200000,
"max_completion_tokens": 16384,
"pricing": {
"input": 0.000003,
"output": 0.000015,
"cache_read": 3e-7
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"extended_thinking"
]
},
{
"id": "anthropic/claude-opus-4.5",
"name": "Claude Opus 4.5",
"context_length": 200000,
"max_completion_tokens": 64000,
"pricing": {
"input": 0.000005,
"output": 0.000025,
"cache_read": 5e-7
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"computer_use"
]
},
{
"id": "openai/gpt-5.4",
"name": "GPT-5.4",
"context_length": 1000000,
"max_completion_tokens": 128000,
"pricing": {
"input": 0.0000025,
"output": 0.000015,
"cache_read": 2.5e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"long_context",
"prompt_caching"
]
},
{
"id": "openai/gpt-5.2",
"name": "GPT-5.2",
"context_length": 400000,
"max_completion_tokens": 128000,
"pricing": {
"input": 0.00000175,
"output": 0.000014,
"cache_read": 1.75e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"prompt_caching"
]
},
{
"id": "openai/gpt-5-mini",
"name": "GPT-5 Mini",
"context_length": 400000,
"max_completion_tokens": 128000,
"pricing": {
"input": 2.5e-7,
"output": 0.000002,
"cache_read": 2.5000000000000002e-8
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"cost_effective",
"prompt_caching"
]
},
{
"id": "openai/gpt-4.1",
"name": "GPT-4.1",
"context_length": 1047576,
"max_completion_tokens": 32768,
"pricing": {
"input": 0.000002,
"output": 0.000008,
"cache_read": 5e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"tools",
"long_context",
"prompt_caching"
]
},
{
"id": "openai/gpt-4.1-mini",
"name": "GPT-4.1 Mini",
"context_length": 1047576,
"max_completion_tokens": 32768,
"pricing": {
"input": 4.0000000000000003e-7,
"output": 0.0000016000000000000001,
"cache_read": 1.0000000000000001e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"tools",
"cost_effective",
"long_context",
"prompt_caching"
]
},
{
"id": "openai/o3",
"name": "o3",
"context_length": 200000,
"max_completion_tokens": 100000,
"pricing": {
"input": 0.000002,
"output": 0.000008,
"cache_read": 5e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"deep_reasoning"
]
},
{
"id": "openai/o4-mini",
"name": "o4-mini",
"context_length": 200000,
"max_completion_tokens": 100000,
"pricing": {
"input": 0.0000011,
"output": 0.0000044,
"cache_read": 2.75e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"cost_effective"
]
},
{
"id": "openai/gpt-audio-1.5",
"name": "gpt-audio-1.5",
"context_length": 128000,
"max_completion_tokens": 16384,
"pricing": {
"input": 0.000004,
"output": 0.000016,
"cache_read": 4.0000000000000003e-7
},
"modality": {
"input": [
"text",
"audio",
"image"
],
"output": [
"text",
"audio"
]
},
"capabilities": [
"vision",
"audio_input",
"audio_output",
"tools",
"multimodal"
]
},
{
"id": "openai/gpt-5-codex",
"name": "GPT-5-Codex",
"context_length": 400000,
"max_completion_tokens": 128000,
"pricing": {
"input": 0.00000125,
"output": 0.00001,
"cache_read": 1.25e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"coding",
"prompt_caching"
]
},
{
"id": "openai/gpt-5.3-codex",
"name": "GPT-5.3-Codex",
"context_length": 400000,
"max_completion_tokens": 128000,
"pricing": {
"input": 0.000005,
"output": 0.00004,
"cache_read": 5e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"coding",
"prompt_caching"
]
},
{
"id": "openai/gpt-realtime-1.5",
"name": "gpt-realtime-1.5",
"context_length": 128000,
"max_completion_tokens": 16384,
"pricing": {
"input": 0.000004,
"output": 0.000016,
"cache_read": 4.0000000000000003e-7
},
"modality": {
"input": [
"text",
"audio",
"image"
],
"output": [
"text",
"audio"
]
},
"capabilities": [
"vision",
"audio_input",
"audio_output",
"tools",
"realtime"
]
},
{
"id": "google/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"context_length": 1000000,
"max_completion_tokens": 65536,
"pricing": {
"input": 0.000002,
"output": 0.000012,
"cache_read": 2.0000000000000002e-7
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"long_context",
"prompt_caching"
]
},
{
"id": "google/gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"context_length": 1000000,
"max_completion_tokens": 65536,
"pricing": {
"input": 5e-7,
"output": 0.000003,
"cache_read": 5.0000000000000004e-8
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"tools",
"long_context",
"cost_effective",
"prompt_caching"
]
},
{
"id": "google/gemini-3.1-flash-lite-preview",
"name": "Gemini 3.1 Flash Lite Preview",
"context_length": 1000000,
"max_completion_tokens": 65536,
"pricing": {
"input": 2.5e-7,
"output": 0.0000015,
"cache_read": 2.5000000000000002e-8
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"tools",
"long_context",
"cost_effective",
"prompt_caching"
]
},
{
"id": "google/gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"context_length": 1048576,
"max_completion_tokens": 65536,
"pricing": {
"input": 0.00000125,
"output": 0.00001,
"cache_read": 1.25e-7
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"long_context",
"thinking",
"prompt_caching"
]
},
{
"id": "google/gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"context_length": 1048576,
"max_completion_tokens": 65536,
"pricing": {
"input": 3e-7,
"output": 0.0000025,
"cache_read": 3e-8
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"long_context",
"thinking",
"prompt_caching"
]
},
{
"id": "google/gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash Lite",
"context_length": 1048576,
"max_completion_tokens": 65536,
"pricing": {
"input": 1.0000000000000001e-7,
"output": 4.0000000000000003e-7,
"cache_read": 1e-8
},
"modality": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"tools",
"long_context",
"cost_effective",
"prompt_caching"
]
},
{
"id": "google/gemini-3.1-flash-image-preview",
"name": "Gemini 3.1 Flash Image Preview",
"context_length": 131072,
"max_completion_tokens": 32768,
"pricing": {
"input": 5e-7,
"output": 0.00006
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"capabilities": [
"vision",
"image_generation",
"multimodal"
]
},
{
"id": "xai/grok-4",
"name": "Grok 4",
"context_length": 256000,
"max_completion_tokens": 65536,
"pricing": {
"input": 0.0000045,
"output": 0.0000225,
"cache_read": 0.000001125
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"deep_reasoning"
]
},
{
"id": "xai/grok-4-1-fast-reasoning",
"name": "Grok 4.1 Fast Reasoning",
"context_length": 2000000,
"max_completion_tokens": 65536,
"pricing": {
"input": 3e-7,
"output": 7.5e-7,
"cache_read": 7.5e-8
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"long_context",
"prompt_caching"
]
},
{
"id": "xai/grok-code-fast-1",
"name": "Grok Code Fast 1",
"context_length": 256000,
"max_completion_tokens": 65536,
"pricing": {
"input": 3e-7,
"output": 0.00000225,
"cache_read": 3e-8
},
"modality": {
"input": [
"text"
],
"output": [
"text"
]
},
"capabilities": [
"reasoning",
"tools",
"coding"
]
},
{
"id": "deepseek/deepseek-v3.2",
"name": "DeepSeek V3.2",
"context_length": 128000,
"max_completion_tokens": 32768,
"pricing": {
"input": 8.000000000000001e-7,
"output": 0.0000024
},
"modality": {
"input": [
"text"
],
"output": [
"text"
]
},
"capabilities": [
"reasoning",
"tools",
"coding",
"cost_effective"
]
},
{
"id": "deepseek/deepseek-r1",
"name": "DeepSeek R1",
"context_length": 128000,
"max_completion_tokens": 32768,
"pricing": {
"input": 0.00000175,
"output": 0.000007
},
"modality": {
"input": [
"text"
],
"output": [
"text"
]
},
"capabilities": [
"reasoning",
"deep_reasoning",
"chain_of_thought",
"coding"
]
},
{
"id": "meta/llama-4-maverick",
"name": "Llama 4 Maverick",
"context_length": 1000000,
"max_completion_tokens": 16384,
"pricing": {
"input": 3e-7,
"output": 0.00000125
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"tools",
"long_context",
"multilingual"
]
},
{
"id": "meta/llama-4-scout",
"name": "Llama 4 Scout",
"context_length": 10000000,
"max_completion_tokens": 16384,
"pricing": {
"input": 2.2e-7,
"output": 8.5e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"tools",
"extreme_context",
"multilingual"
]
},
{
"id": "meta/llama-3.2-90b",
"name": "Llama 3.2 90B Vision",
"context_length": 128000,
"max_completion_tokens": 8192,
"pricing": {
"input": 9.000000000000001e-7,
"output": 9.000000000000001e-7
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"multimodal"
]
},
{
"id": "qwen/qwen3-235b",
"name": "Qwen3 235B",
"context_length": 131072,
"max_completion_tokens": 32768,
"pricing": {
"input": 3e-7,
"output": 0.00000115
},
"modality": {
"input": [
"text"
],
"output": [
"text"
]
},
"capabilities": [
"reasoning",
"tools",
"multilingual",
"thinking"
]
},
{
"id": "qwen/qwen3-coder-480b",
"name": "Qwen3 Coder 480B",
"context_length": 262144,
"max_completion_tokens": 65536,
"pricing": {
"input": 6e-7,
"output": 0.00000235
},
"modality": {
"input": [
"text"
],
"output": [
"text"
]
},
"capabilities": [
"tools",
"coding",
"long_context"
]
},
{
"id": "mistral/mistral-large-3",
"name": "Mistral Large 3",
"context_length": 128000,
"max_completion_tokens": 16384,
"pricing": {
"input": 6.5e-7,
"output": 0.000002
},
"modality": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"capabilities": [
"vision",
"reasoning",
"tools",
"multilingual"
]
}
],
},
],
}1
2
npm install -g openclaw
3
CLI One-Liner
openclaw onboard --non-interactive \
--provider-name=therouter \
--provider-type=openai \
--provider-url=https://api.routerai.lol/v1 \
--provider-api-key=${THEROUTER_API_KEY} \
--default-model=anthropic/claude-sonnet-4.6
OpenClaw Configuration (35 models)
Expand full config ▼
clawhub install therouter-api
npx @skill-hub/cli install therouter-api
Frequently Asked Questions
Common questions about using Router AI with OpenClaw deployments
What is OpenClaw?
Does Router AI work with self-hosted OpenClaw?
How does failover work for channels?
Can I use my own provider API keys (BYOK)?
Which models support vision/audio for channels?
How is usage billed for multi-channel deployments?
Per-token billing with channel-level tracking. The dashboard shows usage breakdown by channel, model, and user. You can export CSVs for reconciliation or integrate with your billing system via the Analytics API. Custom tags in requests allow you to attribute costs to specific channels, users, or campaigns. See the usage accounting guide for details.
Ready to Get Started?
Sign up for Router AI and connect your OpenClaw deployment in minutes. No credit card required to start.
OpenClaw Integration
OpenClaw + Router AI: AI Agents, Every Channel
Power OpenClaw deployments with health-aware routing, multimodal support, and 275+ models. Automatic failover, cost visibility, enterprise security.
Get Started →
Why Router AI for OpenClaw
Build reliable, cost-effective AI agents across every channel
Automatic Failover
Health-aware routing with instant provider switching. If one provider fails, your channels stay online with zero manual intervention.
Multimodal Ready
Vision, audio, PDF, and video support across all channels. Run rich multimedia workflows with WhatsApp, Telegram, Discord, and Slack.
Cost Visibility
Per-channel usage tracking with token-level accounting. See exactly what each bot, channel, and conversation costs in real-time.
Enterprise Security
API key isolation, audit logging, GDPR/CCPA compliance. Zero prompt storage by default. Built for enterprise trust requirements.
Unified API
Single OpenAI-compatible endpoint for 275+ models across Anthropic, OpenAI, Google, Bedrock, xAI. No vendor-specific code.
Direct Connections
Every request hits the real provider's real model. No substitution, no pools, no middlemen. Verify on every single request.
Multimodal Capabilities Across Providers
Run vision, audio, and document workflows across all channels
Anthropic OpenAI Google (Vertex AI) Bedrock xAI
Text
Vision
Audio
PDF
Video
✓ = Full support • = Partial support — = Not supported
Power AI Across Every Channel
Build once, deploy everywhere
WhatsApp
Rich multimedia support for business conversations
Telegram
Fast, secure bots with inline queries and webhooks
Discord
Community engagement with slash commands and threads
Slack
Workplace automation and team collaboration
WeChat
China's leading messaging platform for business and social
Enterprise Security & Trust
Built for production deployments with compliance, audit trails, and privacy guarantees
KMS Envelope Encryption
BYOK credentials protected by AWS KMS with per-credential data keys, AES-256-GCM, and hardware-backed master keys.
Audit Logging
Complete request history with channel, model, cost, and latency tracking.
GDPR/CCPA Ready
Data retention controls, export API, and right-to-deletion support.
Zero Prompt Storage
Request bodies pass through but aren't persisted by default. Metadata only.
View Full Security Details →
Get Started in Minutes
Three steps to production-ready OpenClaw deployment
Sign Up & Get API Key
Create your Router AI account and generate an API key from the dashboard.
Get API Key →
Install OpenClaw
Install the OpenClaw framework globally or in your project.
npm install -g openclaw
Configure & Deploy
Copy-paste the JSON5 config into your OpenClaw setup and you're ready.
CLI One-Liner
Copy
openclaw onboard --non-interactive \
--provider-name=therouter \
--provider-type=openai \
--provider-url=https://api.routerai.lol/v1 \
--provider-api-key=${THEROUTER_API_KEY} \
--default-model=anthropic/claude-sonnet-4.6
OpenClaw Configuration (35 models)
Copy
Expand full config ▼
View full config & models →
Using cc-switch too? Open the cc-switch guide →
Or Install Router AI Skill
Works with Claude Code, OpenClaw, Cursor, Gemini CLI, and 30+ agent tools
clawhub install therouter-api
npx @skill-hub/cli install therouter-api
Frequently Asked Questions
Common questions about using Router AI with OpenClaw deployments
What is OpenClaw?
Does Router AI work with self-hosted OpenClaw?
How does failover work for channels?
Can I use my own provider API keys (BYOK)?
Which models support vision/audio for channels?
How is usage billed for multi-channel deployments?
Per-token billing with channel-level tracking. The dashboard shows usage breakdown by channel, model, and user. You can export CSVs for reconciliation or integrate with your billing system via the Analytics API.
Custom tags in requests allow you to attribute costs to specific channels, users, or campaigns. See the usage accounting guide for details.
Ready to Get Started?
Sign up for Router AI and connect your OpenClaw deployment in minutes. No credit card required to start.
Get Your API Key →
Read the Docs →
Frequently Asked Questions
What is OpenClaw?
OpenClaw is an open-source agent framework for building AI-powered bots across messaging channels like WhatsApp, Telegram, Discord, and Slack. Router AI provides the model gateway layer—handling routing, failover, billing, and multimodal support for your OpenClaw agents.
Does Router AI work with self-hosted OpenClaw?
Yes. Router AI is a cloud gateway you point to from any OpenClaw deployment (cloud or self-hosted). Just configure the baseURL and API key in your OpenClaw config, and you're ready. No additional infrastructure required.
How does failover work for channels?
If a provider fails mid-conversation, Router AI automatically routes to the next healthy provider with the same model. Channel users see no interruption—the conversation continues seamlessly. Health checks run every 30 seconds, and failover happens within milliseconds.
Can I use my own provider API keys (BYOK)?
Yes. Router AI supports Bring Your Own Key (BYOK) for Anthropic, OpenAI, Bedrock, Vertex AI, and xAI. You keep your negotiated pricing, and we handle the routing logic. Your keys are encrypted with AES-256-GCM and never logged. See the BYOK guide for setup instructions.
Which models support vision/audio for channels?
Vision: Claude (Anthropic), GPT-4o (OpenAI), Gemini (Google), and Bedrock models all support image inputs. xAI has partial support. Audio input: GPT-4o (OpenAI) and Gemini (Google) support audio. See the multimodal capabilities matrix above for the full breakdown, or check the multimodal docs.
How is usage billed for multi-channel deployments?
Per-token billing with channel-level tracking. The dashboard shows usage breakdown by channel, model, and user. You can export CSVs for reconciliation or integrate with your billing system via the Analytics API. Custom tags in requests allow you to attribute costs to specific channels, users, or campaigns. See the usage accounting guide for details.