Skip to main content
Velqa

Velqa API

One API key. Two ways to pay.

Connect your existing OpenAI or Anthropic tools to Velqa, then choose the billing model that matches your usage.

Subscription

A predictable monthly quota for regular work. Choose a plan when you want included usage and a clear monthly budget.

Pay-as-you-go

Pay only for what you use with Boost credit. Enable fallback and fund the balance to use it after your included quota.

Start with your existing stack

Your key is managed from the dashboard. The API keeps the OpenAI and Anthropic formats your SDKs already use.

Two compatible endpoints, the same models

Point your client at Velqa by changing the base URL and the key; nothing else in your code moves. Both endpoints share the model catalog and accept the same sk-… key.

OpenAI-compatible

https://api.velqa.dev/v1

For the openai SDK, LangChain and most libraries and coding tools: /v1/chat/completions, streaming and tool calling, with the request format you already use.

Chat API

Anthropic-compatible

https://api.velqa.dev

For tools that speak the Anthropic Messages protocol natively, such as Claude Code or the anthropic SDK: /v1/messages, without an Anthropic account or a local proxy.

Anthropic API

Works with the coding agents you already use

Any OpenAI-compatible client works without a plugin, and Claude Code connects through the Anthropic endpoint. Each guide gives the exact configuration, the context window to set per model and an end-to-end test; the Integrations page of your dashboard generates the snippet ready to paste.

How the two billing modes fit together

What the subscription covers

Your plan sets a monthly token budget, prompt allocations per 5-hour session and per week, and rate limits per model. As long as you stay within them, every request is covered and the Boost balance is not touched.

When a limit is reached

Without fallback, requests are blocked until the next period. With automatic Boost fallback turned on in Billing, the next requests are charged to your prepaid Boost balance instead, so an agent in the middle of a task keeps working.

Pay-as-you-go only

Boost also works on its own: fund the balance and pay per use for the models, including those billed per unit. Boost is prepaid; it creates no subscription and triggers no automatic charge.

How Boost fallback works

One key per tool, revocable on its own

Create a named key for each machine, project or tool. If a laptop is lost or a key leaks into a repository, revoke that one key from the dashboard and the others keep working. The Usage page lists every request with its tokens and cost, so your spend never comes as a surprise.

Manage API keys