---
title: "OpenRouter pricing: the 5.5% fee, free models, October 2026"
canonical: https://uplinkswitch.ai/blog/openrouter-pricing
generated: 2026-10-22T21:17:42.895Z
source: https://uplinkswitch.ai/blog/openrouter-pricing
base_url: https://api.uplinkswitch.com/v1
---

# OpenRouter pricing explained: the 5.5% fee by top-up size, free models and BYOK

OpenRouter pricing in one table: what the 5.5% fee and its $0.80 minimum cost at each top-up size, the free-model limits, the BYOK allowance, and the same $100 here.

October 22, 2026·UplinkSwitch·[All posts](https://uplinkswitch.ai/blog)

OpenRouter pricing has two parts. Tokens bill at each provider's own price, passed through with no markup, and a platform fee lands when you buy credits: 5.5% on the Standard plan with a $0.80 minimum, 8% on Business and 5% on crypto (its [pricing page](https://openrouter.ai/pricing) and [FAQ](https://openrouter.ai/docs/faq), read 2026-10-22). So $100 of credit costs $105.50 on a card, and $10 costs $10.80.

This post works the fee at six top-up sizes, lists the free-model limits and the 18 free ids in OpenRouter's model list on 2026-10-22, prices the bring-your-own-key allowance, and puts the same $100 next to a key priced under list. For OpenRouter against the other gateways, see [OpenRouter and eight alternatives ranked by the fee on top](https://uplinkswitch.ai/alternatives/openrouter).

## How OpenRouter pricing works

OpenRouter sells prepaid credit in US dollars. Each request deducts its token cost from the balance at the model's posted rate, and that rate is the same on every plan. What changes by plan is the fee on the credit, the limits and the team features.

| Plan | Platform fee | Models | Rate limits | Your own provider keys |
| --- | --- | --- | --- | --- |
| Free | None | 25+ free models from 4 free providers | 50 requests a day | No |
| Standard (pay as you go) | 5.5% on credit purchases, $0.80 minimum; 5% on crypto | 500+ models from 80+ providers | Passed through from the providers | $25,000 a month of list-price inference free of fees, then 5% |
| Business (pay as you go) | 8% on credit purchases | 500+ models from 80+ providers | Passed through from the providers | Same as Standard |
| Enterprise | Fee discounts available | 500+ models from 80+ providers | Higher allowances available | Custom |

From openrouter.ai/pricing and its FAQ, read 2026-10-22. The $0.80 minimum and the 5% crypto fee are from the docs FAQ the same day.

Its pricing FAQ answers the markup question directly: inference is billed at the provider's list price on every plan, and the fee is charged when you buy credits, never on individual requests. A failed attempt is not charged for its tokens, so with fallback routing you pay only for the run that answers.

For Claude, GPT and Grok the provider is the developer, so the OpenRouter rate is the price on [Anthropic's pricing page](https://platform.claude.com/docs/en/about-claude/pricing) or [OpenAI's pricing page](https://developers.openai.com/api/docs/pricing). Open-weight models such as DeepSeek and GLM run on several providers, and OpenRouter's model list showed some of them under the developer's own API price on 2026-10-22.

Business costs more per purchase and adds EU and US in-region routing, 1,000 workspaces against 5, and advanced notifications. Enterprise adds invoicing, SSO, contractual SLAs and a dedicated support team. Taxes come on top where they apply.

## The 5.5% fee at each top-up size

| Credit bought | Card fee, Standard | Share of the purchase | You pay by card | Business, 8% | Crypto, 5% |
| --- | --- | --- | --- | --- | --- |
| $10.00 | $0.80 | 8.0% | $10.80 | $10.80 | $10.50 |
| $20.00 | $1.10 | 5.5% | $21.10 | $21.60 | $21.00 |
| $50.00 | $2.75 | 5.5% | $52.75 | $54.00 | $52.50 |
| $100.00 | $5.50 | 5.5% | $105.50 | $108.00 | $105.00 |
| $500.00 | $27.50 | 5.5% | $527.50 | $540.00 | $525.00 |
| $1,000.00 | $55.00 | 5.5% | $1,055.00 | $1,080.00 | $1,050.00 |

Fee rules from OpenRouter's pricing page and FAQ, read 2026-10-22: 5.5% with a $0.80 minimum on Standard, 8% on Business, 5% on crypto. The minimum is shown for card purchases on Standard, where the FAQ states it.

Below $14.55 the $0.80 minimum is larger than 5.5%, so small top-ups pay more. A $10 purchase pays 8.0%, the same rate as Business. From $20 up the card fee is a flat 5.5%.

The fee is part of the purchase, not of the usage, and it does not come back. OpenRouter refunds unused credit within 24 hours of the purchase, keeps the platform fee, and never refunds crypto. Unused credit may expire a year after purchase. Auto top-up buys credit when the balance falls under a threshold you set, and each automatic purchase pays the same fee.

Requesty charges its 5% on each request's model cost instead, and Cloudflare's Unified Billing adds 5% to the credit. The [Requesty alternatives page](https://uplinkswitch.ai/alternatives/requesty) works one coding month on both rules.

## What $100 of OpenRouter credit buys, next to the same $100 here

On OpenRouter, $100 of credit costs $105.50 by card and buys tokens at the developer's list price. On UplinkSwitch, $100 of credit costs $100 and buys tokens at the live catalog rate, under list on 58 of the 75 chat models with a list price.

| Model | OpenRouter, in / out per 1M | UplinkSwitch, in / out per 1M | Under list |
| --- | --- | --- | --- |
| [Claude Opus 5.5](https://uplinkswitch.ai/models/claude-opus-5-5)anthropic/claude-opus-5.5 · claude-opus-5-5 | $4.00 / $20.00 | $2.80 / $10.00 | 30% in, 50% out |
| [Claude Sonnet 5.5](https://uplinkswitch.ai/models/claude-sonnet-5-5)anthropic/claude-sonnet-5.5 · claude-sonnet-5-5 | $2.00 / $10.00 | $1.20 / $3.00 | 40% in, 70% out |
| [Claude Haiku 5.5](https://uplinkswitch.ai/models/claude-haiku-5-5)anthropic/claude-haiku-5.5 · claude-haiku-5-5 | $0.10 / $0.50 | $0.06 / $0.15 | 40% in, 70% out |
| [GPT-5.5](https://uplinkswitch.ai/models/gpt-5.5)openai/gpt-5.5 · gpt-5.5 | $5.00 / $30.00 | $1.80 / $9.00 | 64% in, 70% out |
| [GPT-6.1 Sol](https://uplinkswitch.ai/models/gpt-6.1-sol)openai/gpt-6.1-sol · gpt-6.1-sol | $2.00 / $10.00 | $0.60 / $3.00 | 70% in, 70% out |
| [Grok 4.7](https://uplinkswitch.ai/models/grok-4.7)x-ai/grok-4.7 · grok-4.7 | $2.00 / $6.00 | $0.60 / $1.80 | 70% in, 70% out |

USD per million tokens. OpenRouter: its model list on 2026-10-22, the developer's list price for each of these models, plus 5.5% on the credit that pays for them. UplinkSwitch: GET /v1/catalog when this page was built (2026-10-22).

A steady coding month makes it concrete. 150 turns a day for 20 working days, each turn 80,000 cached, 2,000 input and 1,500 output tokens on Claude Sonnet 5.5, is $81.00 at Anthropic's list. On OpenRouter that needs $85.46 of card purchases; here it is $38.80. The [Claude Code pricing post](https://uplinkswitch.ai/blog/claude-code-api-cost) turns the same shape into light and heavy months.

The model behind each id is checked, not assumed. The [daily real-model check](https://uplinkswitch.ai/proof) sends every model the same prompt and publishes the verdict (230 of 269 stable at the last run), and every non-streamed answer carries a signed receipt that [statements/verify](https://uplinkswitch.ai/statements/verify) checks in the browser. The term-by-term version of this comparison is [UplinkSwitch vs OpenRouter](https://uplinkswitch.ai/vs/openrouter).

## OpenRouter free models: the limits and the list

Free models cost nothing per token and carry their own caps: 20 requests a minute and 50 a day, rising to 1,000 a day once the account has bought $10 of credits. Failed attempts count toward the daily cap, and a negative balance blocks free models until you add credit. The Free Models Router, `openrouter/free`, picks a free model for each request.

The pricing page says 25+ free models. In the public model list on 2026-10-22, 18 ids ended in `:free`:

| Free model id | Developer |
| --- | --- |
| `recursive/flint-v1.0:free` | recursive |
| `minimax/minimax-m3.1-flash:free` | minimax |
| `xingchen-agi/xing4.0-29b-a4b:free` | xingchen-agi |
| `nvidia/nemotron-3.5-super-vl:free` | nvidia |
| `dots-studio/dots-3-note-preview:free` | dots-studio |
| `liquid/lfm-2.5-2.6b:free` | liquid |
| `nvidia/nemotron-3.5-lightning:free` | nvidia |
| `thinkingmachines/inkling-small:free` | thinkingmachines |
| `poolside/laguna-s-2.1:free` | poolside |
| `thinkingmachines/inkling:free` | thinkingmachines |
| `poolside/laguna-xs-2.1:free` | poolside |
| `cohere/north-mini-code:free` | cohere |
| `nvidia/nemotron-3.5-content-safety:free` | nvidia |
| `nvidia/nemotron-3-ultra-550b-a55b:free` | nvidia |
| `nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free` | nvidia |
| `google/gemma-4-26b-a4b-it:free` | google |
| `google/gemma-4-31b-it:free` | google |
| `nvidia/nemotron-3-super-120b-a12b:free` | nvidia |

From https://openrouter.ai/api/v1/models on 2026-10-22: the ids ending in :free, in the order the list returned them. The list changes; read it again before you build on one.

A free id and its paid twin are separate rows in that list. When an app outgrows 50 or 1,000 requests a day it moves to the paid id, at the provider's price plus the credit fee. For roleplay front ends that started on free models, the [OpenRouter alternative for Janitor AI](https://uplinkswitch.ai/blog/openrouter-alternative-janitor-ai) prices 100 messages on the paid builds instead.

## Bring your own key: the $25,000 allowance and the 5% after it

With your own provider keys, the provider bills you for inference directly and OpenRouter charges no fee on the first $25,000 of list-price inference a month, on Standard and Business. Above that it charges 5% of what the same model and provider would cost on OpenRouter, taken from your OpenRouter credits. Enterprise allowances are custom.

A team running $40,000 a month of list-price usage through its own keys pays the providers $40,000, then 5% of the $15,000 over the allowance, which is $750.00 in OpenRouter credits. Under the allowance, routing your own keys through OpenRouter costs the same as calling the providers direct.

Self-hosting a gateway is the other way to keep your own keys with no fee. The [OpenRouter vs LiteLLM comparison](https://uplinkswitch.ai/compare/openrouter-vs-litellm) works out when a server costs less than the 5.5%.

## Is OpenRouter cheaper than going direct?

For Claude, GPT and Grok, no. OpenRouter bills the developer's list price, the same as going direct, then adds 5.5% to the credit. $1,000 of usage at list costs $1,055.00 through OpenRouter and $1,000 at the developer. For open-weight models it can be cheaper, because competing hosts set those prices, and the free models cost nothing up to their caps.

Paying less than direct for the same closed model takes a gateway that prices under the developer's list. On 2026-10-22, 58 of the 75 chat models with a list price sit under it here on input and output, Claude Sonnet 5.5 at $1.20 in and $3.00 out against Anthropic's $2.00 and $10.00. Credit is prepaid from $10 with no fee on the top-up, and each top-up locks your rates for 30 days.

## Estimate a month on OpenRouter

The arithmetic has three steps. Multiply your monthly input and output tokens by each model's rate per million. Add 5.5% for the credit that pays for it, or 8% on Business. Round the purchase up to the top-up sizes you actually buy, since a run of small purchases pays the $0.80 minimum each time.

Cached input lowers the first step. Coding agents re-send the whole conversation on every turn, so most of their tokens bill at the cached rate, and the model pages here list it next to input and output. The [LLM API pricing comparison and calculator](https://uplinkswitch.ai/blog/llm-api-pricing-2026) prices your own token counts at list and here, and the [price index](https://uplinkswitch.ai/pricing-index) has every chat model's rate as JSON and CSV.

## Questions

**Is OpenRouter worth using?** For breadth, yes: 468 ids in its public model list, 500+ models from 80+ providers on its pricing page, free models to try, and per-request provider choice (read 2026-10-22). The cost is the fee: 5.5% on every card purchase of credit, 8% on Business. If you call a handful of Claude or GPT models, the same tokens cost 5.5% more than direct, and a gateway priced under list costs less than either. See [the OpenRouter alternatives compared by fee](https://uplinkswitch.ai/alternatives/openrouter).

**Is it cheaper to use OpenRouter?** Usually not for Claude, GPT or Grok. OpenRouter bills those at the developer's list price, the same as going direct, then adds 5.5% when you buy the credit, so $1,000 of usage costs $1,055.00. It can be cheaper for open-weight models, which it routes to competing hosts, and for the free models. Your own keys within the $25,000 monthly allowance cost the same as direct.

**Do I have to pay for OpenRouter?** Not to start. The Free plan has 25+ free models at 20 requests a minute and 50 a day, and every new user gets a small free allowance (pricing page and FAQ, read 2026-10-22). Paid models need credit, bought with the fee on top.

**How much can I use OpenRouter for free?** 50 free-model requests a day at up to 20 a minute, rising to 1,000 a day once the account has bought $10 of credits. Failed attempts count toward the daily cap, and a negative balance blocks the free models until you add credit (pricing page, read 2026-10-22).

**Is there an OpenRouter promo code?** Its pricing page and FAQ name none (read 2026-10-22). What they list instead: a small free allowance for new users, the free models, $25,000 a month of your own-key usage with no fee, and fee discounts on Enterprise. The FAQ says OpenRouter does not discount inference and offers no volume discounts.

**What does OpenRouter charge on crypto?** 5% of the purchase, against 5.5% with a $0.80 minimum on a card (FAQ, read 2026-10-22). Crypto purchases are never refundable; card purchases can be refunded within 24 hours, without the platform fee.

**Do OpenRouter credits expire?** Its FAQ says OpenRouter reserves the right to expire unused credits one year after purchase (read 2026-10-22). Prepaid credit on UplinkSwitch has no expiry in the terms; top-ups are not refundable. See [the UplinkSwitch terms](https://uplinkswitch.ai/terms).

## Sources

OpenRouter is a trademark of its owner. This post quotes its [pricing page](https://openrouter.ai/pricing), [FAQ](https://openrouter.ai/docs/faq) and [model list](https://openrouter.ai/api/v1/models) as read on 2026-10-22. Our figures come from the live catalog, the daily check at [/proof](https://uplinkswitch.ai/proof) and the [terms](https://uplinkswitch.ai/terms). Get a key: prepaid from $10, no subscription, 30-day price lock, in the [console](https://uplinkswitch.ai/console).

## Related

-   [OpenRouter alternatives by fee](https://uplinkswitch.ai/alternatives/openrouter)
-   [UplinkSwitch vs OpenRouter, one to one](https://uplinkswitch.ai/vs/openrouter)
-   [OpenRouter vs LiteLLM](https://uplinkswitch.ai/compare/openrouter-vs-litellm)
-   [Requesty alternatives](https://uplinkswitch.ai/alternatives/requesty)
-   [OpenRouter uncensored models](https://uplinkswitch.ai/blog/openrouter-uncensored)
-   [OpenRouter alternative for Janitor AI](https://uplinkswitch.ai/blog/openrouter-alternative-janitor-ai)
-   [LLM API pricing comparison, October 2026](https://uplinkswitch.ai/blog/llm-api-pricing-2026)
-   [LLM API price index](https://uplinkswitch.ai/pricing-index)
-   [Daily real-model check](https://uplinkswitch.ai/proof)
-   [Verify a signed statement](https://uplinkswitch.ai/statements/verify)