How much does OpenRouter cost per million tokens?

There is no single OpenRouter token price. The model, provider endpoint, cache behavior, and context tier determine the rate. Use the linked model rows and calculator for a dated text-only estimate, then check your actual route.

Are OpenRouter fees added to every API request?

The platform fee is charged when you purchase credits, rather than as a markup on each model request. Keep the credit purchase payment separate from the inference amount deducted from your balance.

Does BYOK mean there are no OpenRouter charges?

Not always. The provider charges your own account for inference, and OpenRouter can charge a BYOK fee after the applicable monthly allowance. The OpenRouter wallet does not include the provider's separate balance.

Can I apply direct-provider Batch discounts to this table?

No Batch rates are included in these OpenRouter reference rows. A direct-provider Batch discount does not automatically apply to a routed synchronous request. Confirm the endpoint and rate before using a Batch estimate.

Why can a request fail when my account has Credits?

A key's own spending limit, an in-flight spending budget, permissions, or rate limits can block a request independently of the account balance. The OpenRouter guide explains how to distinguish account Credits from key limits and troubleshoot 402 errors.

OpenRouter Pricing & Cost Calculator (Oct 2026)

OpenRouter charges for the model and provider that serve each request. Start with these selected text-model reference prices, estimate token spending, then budget separately for buying credits. Each model name links to its OpenRouter source.

Prices checked on October 15, 2026 against the official pricing page.

OpenRouter prices per million tokens (USD)

Model	Input	Cached input	Output
openai/gpt-6.1-sol	$2.00	$0.10	$10.00
anthropic/claude-sonnet-5.5	$2.00	$0.10	$10.00
meta/muse-wonder	$4.00	$0.20	$20.00

Official OpenRouter pricing

Estimate your OpenRouter cost

Model

Input tokens (millions)

Output tokens (millions)

Estimated cost$40.00

Estimates ordinary short-context input and output at the listed rates. Cache reads/writes, reasoning tokens not included in your output count, tools, routing differences, credit purchase fees, and taxes are excluded. Total monthly tokens do not determine a request's context tier.

Compare models with cache-aware estimates

OpenRouter credit purchase fees

Example: buying $100 of credits. The fee is paid at purchase; the model request later draws from the $100 balance.

Standard · 5.5%

Platform fee: $5.50
Total payment: $105.50

Business · 8%

Platform fee: $8.00
Total payment: $108.00

USD examples exclude taxes and payment-method charges. Check the final checkout amount. BYOK has a separate allowance and fee, explained below.

How OpenRouter billing works

Prices are USD per million tokens, checked on the date above. This is a selected catalog snapshot, not a quote for every endpoint. Check the model's provider list, context tier, privacy settings, and routing before estimating a production workload.
Input, cached input, and output are different billing categories. The calculator assumes ordinary input without caching. Models such as GPT-6.1 Sol and Claude Sonnet 5.5 also charge cache writes; the cached-input column alone does not describe the full cache bill.
For GPT-6.1 Sol, the listed reference rates cover short context. Longer requests can use higher rates. Token counts should include billed reasoning output where applicable; monthly totals can consist of many short requests.
OpenRouter passes through provider inference prices. Its platform fee applies when buying credits: 5.5% on Standard and 8% on Business. Payment processing and applicable taxes can affect the checkout total. The token estimate above is the amount of model usage, not the card charge.
With BYOK, your provider bills inference directly. Standard and Business currently include $25,000 of list-price inference per month without an OpenRouter BYOK fee; above that allowance, a 5% OpenRouter fee is deducted from credits. Enterprise terms are custom. This allowance is measured in inference value, not request count.
Free models have separate availability and rate limits. A paid API budget does not purchase an unlimited subscription. Tool calls, web search, cache storage, media, and specialized modes may add charges beyond the text-token estimate.

Track OpenRouter costs before limits stop requests

Quota Monitor uses an OpenRouter Management Key to read account Credits: total credits minus total usage. This is a shared account balance, not a separate wallet for each API key or project. It does not include the balances of providers billed directly through BYOK. Follow the guide for permissions and verification.

Read the OpenRouter monitoring guide

OpenRouter pricing FAQ

How much does OpenRouter cost per million tokens?: There is no single OpenRouter token price. The model, provider endpoint, cache behavior, and context tier determine the rate. Use the linked model rows and calculator for a dated text-only estimate, then check your actual route.
Are OpenRouter fees added to every API request?: The platform fee is charged when you purchase credits, rather than as a markup on each model request. Keep the credit purchase payment separate from the inference amount deducted from your balance.
Does BYOK mean there are no OpenRouter charges?: Not always. The provider charges your own account for inference, and OpenRouter can charge a BYOK fee after the applicable monthly allowance. The OpenRouter wallet does not include the provider's separate balance.
Can I apply direct-provider Batch discounts to this table?: No Batch rates are included in these OpenRouter reference rows. A direct-provider Batch discount does not automatically apply to a routed synchronous request. Confirm the endpoint and rate before using a Batch estimate.
Why can a request fail when my account has Credits?: A key's own spending limit, an in-flight spending budget, permissions, or rate limits can block a request independently of the account balance. The OpenRouter guide explains how to distinguish account Credits from key limits and troubleshoot 402 errors.

Cost optimization

Batch API Pricing: OpenAI, Claude & Gemini Costs

Compare standard and Batch API costs with a calculator using official model prices. Understand turnaround time, caching, retries, and budget assumptions.

Read article

Cost optimization