01
One gateway to every leading model
Skip separate accounts for DeepSeek, GLM, Qwen, and other providers. Use one account and one API key across leading domestic and global models, and switch models without changing your code.
PAI · Model infrastructure
A clearer way to run models.
PAI Router gives your product one dependable path to models, with usage and cost for every request kept in view.
Router
The model infrastructure layer for stable access, usage and charges.
Available
Work
The Agent work platform for planning, running and continuing meaningful work.
Coming soon
One interface. Fewer moving parts.
Use Responses, Chat Completions or Messages with the same PAI model IDs and Team API key.
Explore the APIPOST /v1/responsestext + image · optional state
{
"model": "claude-fable-5",
"input": "Summarize this update for the team.",
"max_output_tokens": 240,
"stream": true
}Model
claude-fable-5
Protocol
Responses
Request ID
x-request-id
Why PAI
Simpler access. Steadier operations.
Bring model access, intelligent routing, billing, and Team controls together behind one dependable gateway.
02
Intelligent routing with automatic fallback
PAI automatically selects a capable, cost-effective provider. If one service fails, traffic moves to another available route so your product stays online.
03
Clear, transparent billing
See the model, token usage, and cost for every request, itemized and fully traceable. Every charge maps to a specific request.
04
Pay with cryptocurrency
Top up directly with cryptocurrency—no overseas credit card required, removing the friction from cross-border access.
05
Compatible protocols, zero-cost migration
Works with the OpenAI and Anthropic protocols and official SDKs. Change one Base URL and keep the rest of your existing code.
06
Team-based access with clear controls
Manage keys and credits independently for each Team. Set key limits, revoke access at any time, and see exactly who used what.
Transparent economics, from catalog to request.
Review official model token rates before you integrate, then inspect the exact charge after each request.
Claude Fable 5
claude-fable-5
Cached input / M
$1.00
Uncached input / M
$10.00
Output / M
$50.00
Operational clarity is part of the product.
A single Request ID connects ownership, model, measured usage and the final charge.
Ownership
Every request belongs to the Team and API key that initiated it.
Stable model identity
One model ID stays unchanged while Router handles execution.
Measured usage
Input, cached input and output tokens stay distinct.
Clear charges
A request with measured usage produces one precise charge.
Put Router to work.
Open Console to choose a model, create an API key and inspect every request.