Short answer: OpenRouter does not mark up inference, but it does charge a platform fee on credit purchases. At $1,000 of credits, the fee is $55 on Standard or $80 on Business. That is cheap for access to 600+ models and multi-provider routing; it is expensive if you are buying the same model every month at scale and a flat plan fits your caps.
OpenRouter plans and fees
| Plan | Platform fee | Models | Free models | Rate limits | Best for |
|---|---|---|---|---|---|
| Free | N/A | 25+ free models | 25+ | 20 RPM, 50/day; 1,000/day after $10 credits | Trying models |
| Standard | 5.5% on credit purchases | 600+ | 25+ | Provider limits pass through | Indie developers, small teams |
| Business | 8% on credit purchases | 600+ | 25+ | Provider limits pass through; 1,000 workspaces | Teams needing controls and in-region routing |
| Enterprise | Custom | 600+ | 25+ | Higher allowances | SSO, SLAs, private endpoints, invoicing |
Source: openrouter.ai/pricing, verified 19 Nov 2026. Fees, limits and features change; confirm before committing budget.
What the fee costs at different volumes
| Credit purchase | Standard 5.5% | Business 8% | Effective cost |
|---|---|---|---|
| $100 | $5.50 | $8.00 | $105.50 / $108.00 |
| $500 | $27.50 | $40.00 | $527.50 / $540.00 |
| $1,000 | $55.00 | $80.00 | $1,055 / $1,080 |
| $5,000 | $275.00 | $400.00 | $5,275 / $5,400 |
| $25,000 | $1,375 | $2,000 | $26,375 / $27,000 |
The fee is charged when you buy credits, not per request. If you spend $1,000 on inference in a month, the platform fee is $55 or $80, not 5.5% or 8% of each token. Taxes and payment-method fees may be added separately.
What is not marked up
- Inference. OpenRouter says model tokens are billed at provider list price on every plan.
- Failed model attempts. An attempt that errors is not charged for model tokens, so fallback only bills the run that answers.
- Free models. The free catalogue is free, subject to the 20 RPM / 50-per-day cap (1,000/day after $10 in credits).
- Routing through
openrouter/auto. No extra fee; you pay the rate of the model that serves the request.
Auxiliary services that already ran, such as web search or file parsing, may still be billed even when the model call fails. Check the current terms for the services you enable.
BYOK: the $25,000 threshold
Bring-your-own-key changes the math. On Standard and Business, OpenRouter includes $25,000 of list-price inference per month through BYOK with no OpenRouter fee. Above that allowance, it charges 5% of what the same request would have cost at list price, deducted from your credits.
If you already have provider contracts, BYOK can be cheaper than buying credits, especially below the $25K threshold. It also means you manage provider keys, quotas and billing directly. If you do not have those contracts, the standard credit fee is the simpler path.
When OpenRouter is the right choice
- You need a specific model that a smaller catalogue does not host. 600+ models is the main reason to use OpenRouter.
- You want multi-provider routing and fallback. OpenRouter’s core product is routing across providers.
- Your volume is low. At $100–$500/month, the platform fee is small and a flat plan may be more expensive.
- You need BYOK with existing provider contracts. The $25K monthly allowance is generous for many teams.
When a flat plan wins
A flat plan wins when the monthly inference bill is above the plan price and the workload fits the plan’s RPM, concurrency and daily caps. At $1,000/month, the OpenRouter fee alone is $55–$80; a flat plan can be cheaper if the model catalogue covers your workload.
Plugsky’s paid plans are $8–$120/month with published caps and 39 models. The LLM API cost calculator shows the break-even against per-token billing; the cheapest LLM API table shows the list-price baseline. If you are moving an integration, the OpenRouter migration guide covers the base URL, model mapping and cutover checklist.
How to reduce the platform fee
- Use BYOK under the $25K monthly allowance. If you already pay a provider directly, this removes the OpenRouter fee on that inference.
- Negotiate Enterprise. Fee discounts are available on Enterprise contracts, along with SSO, SLAs and invoicing.
- Move stable, high-volume workloads to a flat plan. Keep OpenRouter for long-tail models and routing.
- Buy credits in larger batches only if you will use them. The fee is proportional, so a larger purchase does not reduce the rate on Standard or Business.
- Avoid failed retries. Failed model attempts are not billed for tokens, but retries still consume time and can trigger provider rate limits.
FAQ
Does OpenRouter mark up model prices?
No. OpenRouter says inference is billed at each provider’s list price on every plan. Its revenue comes from a platform fee on credit purchases: 5.5% on Standard and 8% on Business at the time of writing. Failed model attempts are not billed for model tokens.
What does $1,000 of OpenRouter credits cost?
At the current Standard fee, $1,000 of credits costs $1,055. On Business, it costs $1,080. The fee is charged when you buy credits, not on each API request. Taxes and payment-method fees may be added separately.
How does OpenRouter BYOK pricing work?
Bring-your-own-key lets you pay the model provider directly. Standard and Business include $25,000 of list-price inference per month through BYOK with no OpenRouter fee; above that, OpenRouter charges 5% of what the request would have cost at list price. Enterprise allowances are custom.
When is a flat-rate API cheaper than OpenRouter?
A flat plan wins when your monthly inference bill is above the plan price and the workload fits the plan’s RPM, concurrency and daily caps. At low volume, OpenRouter’s fee is small and its 600+ model catalogue is the better fit. Run both numbers at your p95 request size.
Are failed or fallback OpenRouter requests billed?
OpenRouter says an attempt that errors is not charged for model tokens, so with fallback you pay only for the run that answers. Auxiliary services that already ran, such as web search or file parsing, may still be billed.
Compare your OpenRouter bill with a flat plan
Plugsky has 39 models, an OpenAI-compatible API and plans from $8/month. Use the calculator to find the break-even at your volume.
Open the calculator → See plans