One key for your model groups
Use each group’s model name in your tools without managing separate upstream APIs.
An API relay grouped by capability
Small tasks. Why big costs?Use XCoffee. Lower Agent costs.
Give repetitive work and everyday tasks to model groups suited to the job. Reserve higher-cost models for planning and hard problems. Configure delegation in your Agent to spend where it matters.
One API key · OpenAI Chat Completions compatible
Your model for planning and hard problems
Plan. Reason. Decide.
Break down goals, design solutions and tackle hard problems.
Our groups, organized by capability
Code. Organize. Use tools.
Handle frequent small tasks with less unnecessary model spend.
Configure task-specific groups in an Agent that supports delegation. Your Agent splits tasks and chooses which group to call.
01 / Models & setup
Start with the task. Pick the group.
See what each group suits and which models it includes. Configure its model name in your Agent to put that group to work.
No model groups have been published yet. Check the user center before purchasing.
Upstream policies select models within each group. Lists and availability may change; a specific model is not guaranteed on every call.
Three fields. Ready to connect.
Use tools that support a custom OpenAI Chat Completions provider.
Base URL: https://api.xcoff.ee/v1
API Key: YOUR_API_KEY
Model: YOUR_AVAILABLE_MODELUse the personal key from your account · Use a model name listed in your account
02 / Account & pass
Simple setup. Clear usage.
Your key, days remaining and today’s usage, together in your account.
See usage and days remaining
Check shared limits, request details and recorded cache hits.
Renew and keep the same key
Keep your key when renewing, with no need to reconfigure your clients.
03 / FAQ
Before you connect
Will it assign tasks for my Agent automatically?
An Agent with delegation support splits tasks and selects a model group. We forward requests by the model field; upstream policies select the model within a group. Configure task delegation in your Agent or client first.
Will it work with my tools?
Use a client that supports a custom Base URL, API key and model name through OpenAI Chat Completions. Tools that only support Anthropic Messages or OpenAI Responses cannot connect directly at this time.
How is usage counted? What happens at the limit?
Input and output tokens are combined per account across all routes: 100,000,000 tokens per UTC day, 20,000,000 per rolling hour, up to 2 concurrent requests, 30 requests per minute and 300 per hour.
Daily tokens reset at 00:00 UTC. Hourly tokens and request limits use rolling windows. Wait when any limit is reached; renewing or rotating a key does not reset usage. Existing passes follow the limits shown in your account.
Do I need a new key when I renew?
No. Renew your existing pass in your account. An active pass extends from its current expiry; an expired pass restarts from renewal. Your key and client configuration stay the same.