One flat price. A token cap you can see.
Subscribe for a monthly token allowance that resets every cycle, or top up with prepaid credit that never expires. No metered overage, no surprise bills.
Subscribe
A flat monthly price with a token allowance that resets each cycle.
Starter
Just try it.$3 per month
- ~3M tokens / month
- Every model, one endpoint
- Bring any OpenAI-compatible agent
- Community support
Pro
The daily driver.$12 per month
- ~13M tokens / month
- Every model, one endpoint
- Priority routing
- Add prepaid credit anytime
- Email support
Team
Teams & agent fleets.$29 per month
- ~33M tokens / month
- Everything in Pro
- Highest-priority routing
- Priority support
Or top up — credit
One-time prepaid credit. Never expires. Also your overflow valve past a subscription allowance. Bigger packs carry a bonus.
How the allowance works
A subscription sets a flat monthly ceiling on usage, measured in tokens, that resets at the start of every billing cycle. The gateway meters each request against your remaining allowance in real time.
When it's exhausted, further requests are rejected with a 429 Too Many Requests that includes the reset date. Your bill stays exactly what you signed up for — the cap protects you instead of billing you.
Because the models run roughly 10× cheaper than frontier providers, a small monthly price buys a genuinely large token allowance.
What is credit
Credit is prepaid balance you buy once. It never expires and never resets — you spend it down at your own pace, with no recurring charge.
It's also the overflow valve: hit your monthly allowance and, instead of a hard stop, requests draw down any credit you've topped up. No interruption — just prepaid usage until the credit runs out.
Larger packs include bonus credit, and the granted amount is fixed server-side at checkout — you always get exactly what the pack promises.
Frequently asked questions
- What models are included?
- A growing catalog of frontier-grade models is included on every plan, all reached through the same OpenAI-compatible endpoint. Hyphen is a multi-model gateway, not a single-provider reseller — the roster grows over time. Model access is identical across tiers; what changes between tiers is your monthly token allowance and support level.
- How many API keys do I get?
- A subscription includes one key. Buy credit and you can create as many keys as you like — every key draws from the same shared balance, so there's no per-key limit and we never charge by key count.
- How is usage measured?
- In tokens. Every model on the gateway costs the same per token, so a token is the honest, predictable unit — no need to reason about dollars per model. You get a monthly token allowance and can watch it draw down live in the console.
- What happens when I hit my allowance?
- Requests return an HTTP 429 with the reset date in the response. Nothing overflows into a surprise bill — usage pauses until your allowance resets at the start of the next cycle. Want to keep going now? Top up with credit and requests spend straight past the monthly allowance.
- Do credits expire?
- No. Credit is prepaid and never expires. It also acts as the overflow valve — once a subscription's monthly allowance is spent, requests draw down your credit balance instead of stopping. The granted amount for each pack is fixed server-side at checkout.
- Can I use my own coding agent?
- Yes. Hyphen is a drop-in OpenAI-compatible base URL. Claude Code, Codex, Cursor, opencode, Cline, and any OpenAI SDK client work by setting two environment variables.