API gateway
One key. Five tiers. The editor you already use.
Serviqo puts a five-tier model lineup behind a single OpenAI- and Anthropic-compatible endpoint. Point your editor at one base URL, paste the key your reseller issued, and choose a tier by name.
https://api.serviqo.aiNothing to install and nothing to sign up for. The key is the whole credential.
The lineup
Five tiers, one endpointLink to this section
Change tier by changing the model name. Same key, same base URL, no second account. Start at the top of the table and move down only when a job stalls.
| Model | What it is for | Context | Max output | Credit use |
|---|---|---|---|---|
| serviqo-t1Flash | Cheap and quick. Good for small edits, questions and iteration. | 131,072 | 8,192 | 1× |
| serviqo-t2Coder | Purpose-built for code. The sensible default for agent work. | 262,144 | 16,384 | 2.7× |
| serviqo-t3Plus | Strong generalist with a very large context window. | 262,144 | 32,768 | 5× |
| serviqo-t4Pro | High-capability coding and long-horizon agent work. | 262,144 | 32,768 | 18× |
| serviqo-t5Flagship | The heavy one. Reach for it when the others stall. | 262,144 | 65,536 | 56× |
Context and max output are the per-request limits the gateway enforces, in tokens. Credit use is how fast a tier spends a key's budget compared with serviqo-t1 — an estimate at a typical mix of input and output. Real spend depends on the mix of your own traffic.
Setup
Three steps, about two minutesLink to this section
Any tool that speaks the OpenAI or Anthropic API already knows how to talk to Serviqo. You are changing where it points, not how it works.
Point your tool at the base URL
Which of these two you need depends on the client, and mixing them up is the single most common setup failure.
Claude Code — no /v1 on the end, it appends its own path https://api.serviqo.aiCline, Roo Code, Cursor, the OpenAI SDKs, curl — the /v1 matters, and no trailing slash https://api.serviqo.ai/v1Paste your key
Wherever the tool asks for an API key. It travels as
Authorization: Bearer …on every request, which is what an OpenAI-compatible client sends anyway. There is no second secret, no org id and no project id to find.Pick a tier
Use the tier name as the model name —
serviqo-t2is the sensible default for agent work. Switching tier is a change of model name and nothing else, so you can do it mid-session without losing the conversation.
The whole thing, for Claude Code
Paste this into a terminal, replacing YOUR_SERVIQO_KEY with your key, then run claude. It applies to that terminal only.
export ANTHROPIC_BASE_URL=https://api.serviqo.ai
export ANTHROPIC_AUTH_TOKEN=YOUR_SERVIQO_KEY
export CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1The docs cover the same for Cline, Roo Code and Cursor, how to make it permanent, and what to check when a request comes back 401.
Keys
What a key controlsLink to this section
The limits live on the key, not on an account behind it. Whoever issued yours set each of these, and you can read all of them back at any time.
- Model access
- A key is scoped to a set of tiers. It sees the ones it was issued for and nothing else — asking for another is refused, never quietly downgraded to a cheaper one.
- Spend budget
- A hard stop, counted in credits. When the budget is gone the gateway refuses the next request rather than letting it run on. It can be set to reset on a period.
- Rate limits
- Requests per minute and tokens per minute, enforced on the key itself. An agent stuck in a loop hits the limit rather than the budget.
- Expiry
- Optional. A key can be issued to stop working on a date, which is what makes a trial or a short contract safe to hand out.
Check a key whenever you like
Paste a key into the key checker and it reports its own live state — tiers, credits used against the budget, rate limits and expiry.
The check is made with the key as its own credential, so a key can only ever ask about itself. Nothing is written down at our end: no account, no session, no record that you looked.
Works with
The tools you already have openLink to this section
No extension of ours, no fork of anything, no SDK to install.
Claude CodeTerminal
Three environment variables. The tier list is read from the gateway, so the model picker shows exactly what your key can use.
ClineVS Code
Add an OpenAI-compatible provider. Everything routes through Serviqo — planning, edits and all.
Roo CodeVS Code
The same provider settings as Cline, with a different tier pinned to each mode if you want one.
CursorChat and Agent
Override the OpenAI base URL and add the tier names by hand; Cursor does not ask the gateway what your key can use.
Anything OpenAI-compatiblecurl, SDKs, your own scripts
Point it at the /v1 URL, send the key as a bearer token, and name a tier as the model. No SDK of ours to install.
One caveat before you start on Cursor: Tab autocomplete stays on Cursor's own servers whatever you configure, and the Cursor setup notes set out exactly which half of the editor that leaves on Serviqo.
Getting a key
Keys are issued by resellersLink to this section
Serviqo does not sell keys. Access is resold, and the reseller you buy from is the one who issues your key and sets what it can do — which tiers it reaches, how much budget it carries, how fast it may be used and when it stops working. That is why there is no pricing page here, no checkout and no account to create.
If you have a key already, everything on this site works for you right now: set up your editor or check what the key can do. If you do not have one yet, it comes from the reseller you buy access through — we cannot issue one from this page.
To top up, raise a limit or extend an expiry, go back to whoever gave you the key. They hold the controls; this site only reads them.