OpenRouter Pricing: Fees, Free Model Limits and a Cheaper Option
OpenRouter pricing explained: list-price tokens, the 5.5% credit fee, free model daily limits, credit expiry and refunds, and where reAPI charges less per token.
OpenRouter pricing has two parts. Every model is billed per token at the price its provider lists, with no markup on the request itself, and OpenRouter takes its cut when you buy credits: a 5.5% platform fee on the Standard plan and 8% on Business[1]. Free models cost nothing but are capped per day, and the cap depends on whether you have bought credits[2].
We read OpenRouter's pricing page, its limits documentation and its public model list on October 5, 2026, and set the per-token prices next to reAPI's model pages for the models both sell. reAPI is our product, so read the reAPI column knowing the source. Where OpenRouter is the better deal, we say so below.
TL;DR
- Token prices: OpenRouter bills each model at its provider's list price, for example $2 input and $10 output per million tokens for GPT-6 Sol[1][3].
- Fees: buying credits costs 5.5% extra on Standard and 8% on Business. Requests that error are not charged[1].
- Free models: 20 requests a minute, and 50 requests a day until you have bought at least $10 of credits, after which the daily cap is 1,000[2].
- Credits: OpenRouter may expire unused credits a year after purchase, and refunds for unused credits must be requested within 24 hours[4].
- Where reAPI costs less: GPT-6 Astra, GPT-6 Sol, GPT-6 Luna, Claude Opus 5.5 and Claude Fable 5.1 list 20% lower per token on reAPI, with no fee on top[3][5][6].
How OpenRouter pricing works
OpenRouter is a router: one API key and one endpoint for hundreds of models from many providers. Its pricing page says inference is billed at each provider's list price on every plan, and that the fee is "charged when you buy credits, 5.5% on Standard and 8% on Business, never on individual requests"[1]. In practice, $100 of credits costs $105.50 on Standard.
| Plan | Platform fee on credits | Payment |
|---|---|---|
| Free | None (free models only) | None |
| Standard | 5.5% | Card, crypto and more |
| Business | 8% | Card, crypto and more |
| Enterprise | Custom, with fee discounts | Invoicing |
Standard and Business have no minimum spend. If you bring your own provider keys (BYOK), the first $25,000 of list-price inference a month carries no OpenRouter fee, and usage beyond that is charged 5%[1]. OpenRouter says it does not discount inference itself; discounts apply only to its fees, on enterprise plans[1].
OpenRouter free models and their limits
Models with a :free variant cost nothing, and the free models router picks one for you. They are rate limited: 20 requests a minute, 50 requests a day on an account that has bought less than $10 of credits, and 1,000 a day once you have bought at least $10[2]. Over the limit, the API returns a too-many-requests error. Free models are useful for testing, but a production workload on them depends on those caps.
OpenRouter prices compared with reAPI
The table uses OpenRouter's public model list and reAPI's model pages, both on October 5, 2026, in dollars per million tokens. For the GPT-6 models, reAPI's figures are its rates for prompts up to 272K tokens[3][5][6].
| Model | OpenRouter input / output | reAPI input / output | Difference |
|---|---|---|---|
| GPT-6 Astra | $10 / $50 | $8 / $40 | −20% / −20% |
| GPT-6 Sol | $2 / $10 | $1.60 / $8 | −20% / −20% |
| GPT-6 Luna | $0.10 / $0.50 | $0.08 / $0.40 | −20% / −20% |
| Claude Opus 5.5 | $4 / $20 | $3.20 / $16 | −20% / −20% |
| Claude Fable 5.1 | $10 / $50 | $8 / $40 | −20% / −20% |
Because OpenRouter's 5.5% fee sits on top of its token prices, the effective gap on Standard is about 24%. A workload of 100 million input and 20 million output tokens on GPT-6 Sol costs $400 in tokens on OpenRouter, $422 once the credit fee is included, and $320 on reAPI[1][5].
Two cases go the other way. reAPI charges a higher long-context rate when a GPT-6 prompt exceeds 272K tokens, for example $3.20 input and $12 output for GPT-6 Sol, while OpenRouter's model list shows one rate for these models, so compare both for very long prompts[3][5]. And DeepSeek V4.1 Flash listed lower on OpenRouter than on reAPI on October 5, 2026, so it is not in the table[3][8].
Moving from OpenRouter to reAPI
Both use the OpenAI chat completions format. OpenRouter's base URL is https://openrouter.ai/api/v1; reAPI's is https://reapi.ai/api/v1, with model IDs such as gpt-6-sol and claude-opus-5-5 instead of OpenRouter's openai/gpt-6-sol style[5][7]. Change the base URL, the key and the model name, then run the same prompts on both and compare the bills. reAPI's catalog is much smaller than OpenRouter's, so check that every model you use is on its models page first.
Frequently asked questions
Is OpenRouter API free?
Only its free models are, within the daily limits above. Every other model is paid per token from prepaid credits, and buying credits adds the 5.5% fee on Standard[1][2].
How much does it cost to use OpenRouter?
The provider's list price per token for each model, plus 5.5% on the credits you buy (8% on Business). There is no subscription and no minimum spend on Standard[1].
What are credits in OpenRouter?
Credits are the prepaid dollar balance your requests draw from. You buy them manually or with auto top-up, each request deducts its cost, and unused credits may expire one year after purchase[1][4].
What is the daily limit for free OpenRouter?
50 requests a day for free models if you have bought less than $10 of credits, and 1,000 a day once you have bought at least $10, with a limit of 20 requests a minute in both cases[2].
Is there a promo code for OpenRouter?
OpenRouter's pricing page says it does not discount inference, and fee discounts are offered on its enterprise plan[1]. We found no official promo code on its public pages.
Is OpenRouter safe?
OpenRouter says it logs request metadata such as timestamps, model and token counts, and does not log prompts or completions unless you opt in[4]. Each request is still processed by the model provider it routes to, so check that provider's data policy for sensitive work.
Choosing between OpenRouter pricing and reAPI
OpenRouter is the better fit if you need its very large catalog, free models for testing, or provider routing across hundreds of models. For GPT-6 Astra, GPT-6 Sol, GPT-6 Luna, Claude Opus 5.5 and Claude Fable 5.1 at normal prompt lengths, reAPI listed 20% lower per-token prices on October 5, 2026, with no fee on credit purchases. If those models are most of your bill, move them and keep OpenRouter for the rest.
References
- OpenRouter. Pricing (plans, platform fees and billing FAQ). Retrieved October 2026 from openrouter.ai/pricing
- OpenRouter. API reference: Limits. Retrieved October 2026 from openrouter.ai/docs/api-reference/limits
- OpenRouter. Models API (public model list with pricing). Retrieved October 2026 from openrouter.ai/api/v1/models
- OpenRouter. FAQ. Retrieved October 2026 from openrouter.ai/docs/faq
- reAPI. GPT-6 Sol, GPT-6 Astra and GPT-6 Luna model pages. Retrieved October 2026 from reapi.ai/models/gpt-6-sol, reapi.ai/models/gpt-6-astra and reapi.ai/models/gpt-6-luna
- reAPI. Claude Opus 5.5 and Claude Fable 5.1 model pages. Retrieved October 2026 from reapi.ai/models/claude-opus-5-5 and reapi.ai/models/claude-fable-5-1
- OpenRouter. Quickstart. Retrieved October 2026 from openrouter.ai/docs/quickstart
- reAPI. DeepSeek V4.1 Flash model page. Retrieved October 2026 from reapi.ai/models/deepseek-v4-1-flash
Author
Categories
More Posts
Is Seedance 2.5 Free? Trials, Credits, and Plan Limits
Is Seedance 2.5 free? Compare conditional trial credits, paid API activation, and unlimited-plan limits without confusing old promotions with current access.
AI Image API Content Filters: How Refusals Happen
Image API filtering runs in several layers. The same prompt can pass one host and fail another, but certain boundaries never change regardless of configuration.
Krea AI Pricing: Plans, Compute Units, API Costs and Is It Free?
Krea AI pricing explained: Free, Basic, Pro and Max plans, how compute units work, the separate Krea API price list, refunds, and where reAPI costs less per request.